Journal of Personalized Medicine

Article ETS-Domain Elk-1 Regulates Stemness in Brain Tumors and CD133+ BrainTumor-Initiating Cells

Melis Savasan Sogut 1,2,3, Chitra Venugopal 4, Basak Kandemir 1,2,3,†, Ugur Dag 3,‡, Sujeivan Mahendram 4, Sheila Singh 4, Gizem Gulfidan 5 , Kazim Yalcin Arga 5 , Bayram Yilmaz 6,* and Isil Aksan Kurnaz 1,2,*

1 Institute of Biotechnology, Gebze Technical University, 41400 Kocaeli, Turkey; [email protected] (M.S.S.); [email protected] (B.K.) 2 Molecular Neurobiology Laboratory (AxanLab), Department of Molecular Biology and Genetics, Gebze Technical University, 41400 Kocaeli, Turkey 3 Biotechnology Graduate Program, Graduate School of Sciences, Yeditepe University, 26 Agustos Yerlesimi, Kayisdagi, 34755 Istanbul, Turkey; [email protected] 4 Stem Cell and Cancer Research Institute, McMaster University, Hamilton, ON L8S 4K1, Canada; [email protected] (C.V.); [email protected] (S.M.); [email protected] (S.S.) 5 Department of Bioengineering, Marmara University, 34722 Istanbul, Turkey; gizemgulfi[email protected] (G.G.); [email protected] (K.Y.A.) 6 Department of Physiology, Faculty of Medicine, Yeditepe University, 26 Agustos Yerlesimi, Kayisdagi, 34755 Istanbul, Turkey * Correspondence: [email protected] (B.Y.); [email protected] (I.A.K.) † Present address: Department of Molecular Biology and Genetics, Baskent University, 06790 Ankara, Turkey. ‡ Present address: Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology,  Cambridge, MA 02139-4307, USA. 

Citation: Sogut, M.S.; Venugopal, C.; Abstract: Elk-1, a member of the ternary complex factors (TCFs) within the ETS (E26 transformation- Kandemir, B.; Dag, U.; Mahendram, specific) domain superfamily, is a transcription factor implicated in neuroprotection, neurodegenera- S.; Singh, S.; Gulfidan, G.; Arga, K.Y.; tion, and brain tumor proliferation. Except for known targets, c-fos and egr-1, few targets of Elk-1 Yilmaz, B.; Kurnaz, I.A. ETS-Domain have been identified. Interestingly, SMN, SOD1, and PSEN1 promoters were shown to be regulated Transcription Factor Elk-1 Regulates by Elk-1. On the other hand, Elk-1 was shown to regulate the CD133 , which is highly expressed Stemness Genes in Brain Tumors and in brain-tumor-initiating cells (BTICs) and used as a marker for separating this cancer stem cell CD133+ BrainTumor-Initiating Cells. population. In this study, we have carried out microarray analysis in SH-SY5Y cells overexpressing J. Pers. Med. 2021, 11, 125. https:// Elk-1-VP16, which has revealed a large number of genes significantly regulated by Elk-1 that function doi.org/10.3390/jpm11020125 in nervous system development, embryonic development, pluripotency, apoptosis, survival, and Nanog Academic Editor: Chiara Villa proliferation. Among these, we have shown that genes related to pluripotency, such as , , and Oct4, were indeed regulated by Elk-1, and in the context of brain tumors, we further showed that

Received: 20 January 2021 Elk-1 overexpression in CD133+ BTIC population results in the upregulation of these genes. When Accepted: 9 February 2021 Elk-1 expression is silenced, the expression of these stemness genes is decreased. We propose that Published: 14 February 2021 Elk-1 is a transcription factor upstream of these genes, regulating the self-renewal of CD133+ BTICs.

Publisher’s Note: MDPI stays neutral Keywords: ETS; Elk-1; stem cell; microarray; brain-tumor-initiating cell (BTIC) with regard to jurisdictional claims in published maps and institutional affil- iations. 1. Introduction The ternary complex factor (TCF) Elk-1 of the ETS domain superfamily is a ubiquitous transcription factor, yet it interacts with neuronal microtubules and motor , is Copyright: © 2021 by the authors. found mainly in neuronal axons and dendrites, and is phosphorylated at Serine 383 residue Licensee MDPI, Basel, Switzerland. in fear conditioning or synaptic plasticity paradigms [1–6]. Phosphorylation of Elk-1 by This article is an open access article MAPKs, in particular Serine 383 and Serine 389 within the activation domain, was shown distributed under the terms and to induce its binding to DNA [7,8]. conditions of the Creative Commons Elk-1 transcription factor has been widely studied with respect to its mitogen-induced Attribution (CC BY) license (https:// activation through phosphorylation by mitogen-activated kinases (MAPKs) and creativecommons.org/licenses/by/ regulation of the c-fos promoter in complex with (SRF) [9]. However, 4.0/).

J. Pers. Med. 2021, 11, 125. https://doi.org/10.3390/jpm11020125 https://www.mdpi.com/journal/jpm J. Pers. Med. 2021, 11, 125 2 of 28

Elk-1 and other ternary complex factor (TCF) members have a rather large number of targets, some of which have a high degree of redundancy [10]. A thousand new promoters were identified for Elk-1 binding using a ChIP-chip assay, with two distinct binding modes: SRF-dependent and SRF-independent; furthermore, it was shown that there was a redundancy of promoter occupancy by other ETS proteins in a subset of promoters [10]. Elk-1 was also shown to regulate survival in neuronal cell models by regulating the Survival of Motor Neuron (SMN) promoter as a novel target [11]. CD133, a widely-accepted Cancer Stem Cell (CSC) marker [12–14], was also shown to be regulated through ets motifs as well as hypoxia-inducible elements, through the interaction of HIF-1α and Elk-1 on the promoter [15]. Elk-1 was recently found to have both activating and repressive role in human embry- onic stem cells (hESCs), particularly through SRF interaction, and found to be upregulated in mesoderm differentiation [16]. In this study, we have first aimed to identify novel targets of Elk-1 using SH-SY5Y neuroblastoma cell line in a transcriptomics approach. We have identified novel pathways and genes that were up- or downregulated upon Elk-1-VP16 overexpression, and when promoters of a subset of these genes were analyzed, several ets motifs were identified. Among these, genes related to pluripotency or early neuronal development were particu- larly interesting, hence we have further analyzed and verified the regulation of a selected set of genes by Elk-1 using qPCR and investigated the regulation of SOX2, NANOG, and POU5F1 promoters by Elk-1 and its binding to predicted ets motifs in neuroblastoma and glioblastoma (GBM) cell lines. Considering Elk-1 was previously shown to regulate CD133 expression [15], we have also studied Elk-1 expression levels in CD133− and CD133+ cell lines as well as primary brain tumors, indicating Elk-1 was indeed overexpressed in CD133+ cells, and when Elk-1 expression was silenced by RNAi, SOX2, and NANOG expression were reduced in both CD133+ primary GBMs, as well as CD133+ cell lines in a cell context-dependent manner.

2. Materials and Methods 2.1. Cell Culture and BTIC Isolation from Cell Lines and Primary Tumors SK-N-BE (2) (ATCC CRL-2271) and SH-SY5Y (ATCC CRL-2266) human neuroblastoma cell lines as well as U-87 MG (ATCC® HTB-14), A172 (ATCC CRL-1620), and T98G (ATCC CRL-1690) human GBM cell lines were used. U87-MG, A172, and T98G cell lines were provided by Assist. Prof. Tugba Bagci Onder from Koc University. For all the stated cell lines for monolayer culture, DMEM high-glucose (4.5 g/L) medium (Gibco, #41966029, Waltham, MA, USA) was used as a basal medium and supplemented with one percent penicillin-streptomycin solution (Gibco, #15140122, Rockville, MA, USA) and 10 percent fetal bovine serum (FBS) (Life Technologies, #10500064, Carlsbad, CA, USA). Cells were grown in 37 ◦C and 5 percent CO2 incubator. To form tumorsphere cultures from monolayer cells and support brain-tumor-initiating cells after conducting CD133+ isolation, initial proliferation media (IPM), N2 media, and coated culture plates were used. Plates were prepared by coating with poly-HEMA (poly (2-hydroxyethyl methacrylate) solution. To prepare poly-HEMA solution, 38 mL absolute ethanol was mixed with two mL double distilled water. Following the addition of 1.2 g of poly-HEMA (Sigma Aldrich, #P3932, Taufkirchen, Germany) powder into the mixture, it was placed in a shaker at 37 ◦C with a vigorous shake for four-five hours until no pow- der could be seen with the naked eye. This poly-HEMA solution was filtered through a 0.22-micron filter and kept at 4 ◦C up to six months. Initial proliferation medium (IPM) is necessary for culturing tumorspheres and isolated brain-tumor-initiating cells up to three passages. IPM is made up of neurobasal medium (Gibco, #21103049, Waltham, MA, USA), 1X B27 (Gibco, #17504044, Waltham, MA, USA), 1X GlutaMAX (Gibco, #35050061, Waltham, MA, USA), one percent penicillin-streptomycin solution (Gibco, #15140122, Waltham, MA, USA), 20 ng/mL FGF-2 (Gibco, #13256029, Waltham, MA, USA), and 20 ng/mL EGF (Gibco, #SRP3027, Waltham, MA, USA). N2 medium is necessary for culturing spheroids and iso- J. Pers. Med. 2021, 11, 125 3 of 28

lated brain-tumor-initiating cells over three passages. N2 medium is made up of neurobasal medium (Gibco, #21103049), 1X N2 (Gibco, #17502048, Waltham, MA, USA), 1X GlutaMAX (Gibco, 35050061, Waltham, MA, USA), one percent penicillin-streptomycin solution (Gibco, #15140122, Waltham, MA, USA), 20 ng/mL FGF-2 (Gibco, #13256029, Waltham, MA, USA), and 20 ng/mL EGF (Waltham, MA, USA, #SRP3027, Waltham, MA, USA). For brain-tumor-initiating cells’ (BTICs) isolation from cell lines, SK-N-BE (2) neurob- lastoma cells were grown as monolayer cells up to 80 percent confluency, and on the day of isolation, the media was removed, cells were washed with five mL PBS/flask, and three mL of StemPro Accutase/flask was added onto the cells. The suspension was centrifuged at 300× g for five minutes. The cells were resuspended with MACS buffer [two percent bovine serum albumin (BSA), two mM EDTA, and phosphate-buffered saline (PBS) pH 7.2]. To prevent the clogging of the columns at the ongoing isolation procedure, cells were passed through the first 70-micron cell strainer several times until they could pass freely through it. Then, they were passed through a 30-micron filter several times, cell aggregates were removed, and the single-cell suspension was prepared. Cells could be counted at this stage of the procedure. The viable cell number was determined by staining the cells with 0.4 percent Trypan Blue Solution (Gibco, #15250061, Waltham, MA, USA). To continue, cells were centrifuged at 300× g for 10 min, and the supernatant was removed. Cells were resuspended in 60 µL MACS buffer/107 cells and 20 µL FcR Blocking Agent/107 cells, and 20 µL CD133 Microbeads/107 cells were added (CD133 MicroBead Kit—Tumor Tissue, human, Miltenyl Biotec, #130-100-857, Gladbach, Germany). The cells were incubated at 4 ◦C for 30 min at a constant, slow rotation (12 rpm). Following incubation, two mL buffer/107 cells were added to wash the cells, and then, they were centrifuged at 300× g for 10 min again. The supernatant was aspirated, and the pellet was resuspended in 500 µL MACS buffer/107 cells and continued with magnetic separation part. MACS MS column (Miltenyl Biotec, #130-042-201, Germany) was placed on the MACS Mini Separation stand and was equilibrated with 500 µL MACS buffer. The cells prepared in the previous step were loaded onto that column, and with gravity effect, the suspended cells flow through the column for positive selection. That is, the cells labeled for CD133 (CD133+) were kept in the column, while marker-free cells (CD133−) would not bind to the column and were collected in a tube. The column was washed three times with 500 µL of the buffer to wash column-retaining CD133+ cells, the flowing liquid was collected again, and the resulting cells were combined to assemble CD133− cells. For elution of CD133+ cells, the column was separated from the magnetic stand and allowed to stand in the non-magnetic field for about two minutes and flushed out with one mL MACS buffer with the supplied plunger. Cells were counted with Trypan Blue, centrifuged for five minutes at 150× g and resuspended in complete IPM and cultured for 7–10 days in a humidified ◦ incubator at 37 C and five percent CO2, replacing the medium with freshly prepared IPM every three–four days until their size reached 200 microns, or they started dying from the center. When they reached the limitations, they were passaged. For passaging the cells, the suspension cells were collected from the dishes to a falcon and centrifuged at 300× g for 10 min. Following the centrifugation, the medium was aspirated, and one mL StemPro Accutase Cell Dissociation Reagent (Gibco, #A1110501, Waltham, MA, USA) was added onto the cells, and cells were incubated at 37 ◦C for five minutes. Cells were triturated about 40 times until the spheroids become single-cell suspension. Onto this single-cell suspension, five mL of PBS with antibiotics was added. Cells were counted at this stage if necessary or to continue cells were centrifuged at 300× g for five minutes. The cells were resuspended in complete IPM at the proper volume.

2.2. Dissociation and Culture of Primary GBM Tissue Human GBM samples were obtained from consenting patients, as approved by the Hamilton Health Sciences/McMaster Health Sciences Research Ethics Board. Brain tu- mor samples were dissociated as previously described [17] and cultured as neurospheres in Neurocult complete (NCC) media, a chemically defined serum-free neural stem cell J. Pers. Med. 2021, 11, 125 4 of 28

medium (STEMCELL Technologies, Vancouver, BC, Canada), supplemented with human recombinant epidermal growth factor (20 ng/mL: STEMCELL Technologies, Vancouver, Canada), basic fibroblast growth factor (20 ng/mL; STEMCELL Technologies, Vancouver, Canada), heparin (2 µg/mL 0.2% Heparin Sodium Salt in PBS; STEMCELL Technologies, Vancouver, Canada), antibiotic-antimycotic (10 mg/mL; Wisent Bioproducts, Saint Bruno, QC, Canada) in ultra-low attachment plates (Corning, New York, NY, USA). Primary GBM cells (BT 428, BT 458 and BT 624) were cultured in NSC complete media and flow-sorted for CD133+ and CD133− populations as described previously [18,19]. Transfections were carried out by Lipofectamine 2000 as per the manufacturer’s instructions.

2.3. Transient Transfection of Cells For transfection of adherent cells, single-cell suspensions of adherent cell cultures were prepared and seeded at 0.3–0.6 × 106 cells/cm2 density in complete DMEM medium, ◦ and they were incubated in 37 C, five percent CO2 incubator, so that they would be 85–90 percent confluent at the time of transfection. On day one, for the formation of the carrier liposome complex, the desired plasmid and PEI were mixed at the determined ratio for each cell line in serum-free DMEM and incubated at room temperature for 20 min. At the end of the period, a complete DMEM medium with 10 percent FBS was added to the mixture at half the volume of the mix. Two hours later, complete DMEM medium containing 10 percent FBS was added to the wells/dishes and the cells were incubated ◦ for 48 h in 37 C, five percent CO2 incubator for the transgene expression. Cells were transfected with empty pCDNA3 or pCMV plasmids, pCMV-Elk-1 and pRSV-Elk-1-VP16 (courtesy of Prof. A.D. Sharrocks) using the PEI reagent (CellnTech), in 3 replicas per sample. psiSTRIKE hMGFP-scrRNA (from here on referred to as scrRNA) and psiSTRIKE hMGFP-siElk-1 (from here on referred to as siElk-1) has been described elsewhere [11]. For transfection of BTICs, the suspension cells were collected from the dishes to a falcon and centrifuged at 300× g for 10 min. Following the centrifugation, the medium was aspirated, and one mL StemPro Accutase Cell Dissociation Reagent (Gibco, #A1110501) was added onto the cells, and cells were incubated 37 ◦C for five minutes. Cells were triturated for about 40 times until the spheroids become single-cell suspension. Onto this single-cell suspension, five mL of PBS with antibiotics was added and centrifuged at 300× g for five minutes. The cells were resuspended in complete IPM at the proper volume. Cell density and Lipofectamine 2000 (Thermo Fischer Scientific, Waltham, MA, USA) ratio were determined. Cells were seeded at 0.3–0.6 × 106 cells/cm2 density in complete IPM without antibiotics on the day of transfection. Following the cell seeding, Lipofectamine 2000 and the nucleic acids were diluted in neurobasal medium without antibiotics, incubated at room temperature for five minutes, then the diluted Lipofectamine 2000 was gently combined with the dilute nucleic acids, and the mixture was incubated at room temperature for 20 min to form liposome. Then, the mixture was added directly onto wells containing cells, and the ◦ cells were incubated for 24–72 h in 37 C, 5% CO2 incubator for the transgene expression

2.4. Soft Agar Assay For softy agar assay, 100 cells in 100 µL IPM and an equal volume of 2.8% low- melting-point (LMP) agarose solution were mixed to generate 1.4% agarose-cell solution ◦ per well in a 96-well plate, and the mixture was incubated at 37 C, 5% CO2 incubator for 14 days. At the end of 14 days, colonies were counted under a 10× magnification or stereo microscope. For staining, crystal violet was dissolved in PBS with two percent ethanol at a final concentration of 0.04 percent, filtered with 0.45 µm filter, and dishes were stained with 50 µL of this solution for one hour at room temperature. The plates were checked every ten minutes to prevent the staining of the background. Then, the staining solution was removed carefully, and the wells were washed with water three times for 30 min. At the last wash, water was kept in the wells overnight to remove the background. The assay was performed in quadruplicate; colonies ≥ 20 µm were counted and analyzed using MS Excel software; results were reported as mean ± standard deviation. J. Pers. Med. 2021, 11, 125 5 of 28

2.5. Limiting Dilution Analysis (LDA) Limiting dilution analysis (LDA) has been extensively used to find out differences within multiple groups for a particular trait. In our case, LDA was used for determining the cancer cell initiating frequency of CD133+ and CD133− SKNBE (2) cells; in other words, to evaluate the self-renewing capacity of BTICs. For LDA, following the BTIC isolation procedure, cells were counted so that 10,000 cells/50 µL complete IPM would be present in the first tube. Through serial dilution by factor two up to 1 cell/50 µL, cells were seeded on poly-HEMA coated 96-well plates. For each condition/cell number, samples were seeded in quintuplet. Twenty-five microliters of culture media were added to each well every three–four days, and cells were examined for the presence/absence of spheres and quantified on day 10.

2.6. RNA Isolation, cDNA Synthesis, Reverse Transcription Polymerase Chain Reaction (RT-PCR), and Real-Time PCR PureLink RNA Mini Kit (Life Technologies Ambion, #12183-018) and PicoPure RNA Isolation Kit (Arcturus, #KIT0202) were used for RNA isolation throughout the experiments. In summary, adherent cells grown in cell culture plates (usually 1.5x106 cells/10 cm culture dish) were washed with cold PBS; then, the resuspended cells were centrifuged at 300× g for 5 min at 4 ◦C. An amount of 0.3–0.6 mL of lysis solution with beta-mercaptoethanol was added onto the cells depending on the number of cells, and they were mechanically burst and homogenized by triturating through an insulin syringe 15 times. The cells were centrifuged at 2000× g for five minutes at 4 ◦C, followed by the addition of 70 percent ethanol equal to the volume of the present cell lysate. The lysates were transferred to the filter cartridges and were centrifuged at 12,000× g for 30 s. This step was repeated until the whole sample was finished, and the washing process was started. For washing, 700 µL of wash buffer I was added and centrifuged at 12,000× g for 15 s. Following the first washing step, 500 µL of wash buffer II was added and repeated twice after centrifugation at 12,000× g for 30 s. Tubes were centrifuged for two minutes to dry the membrane. In the elution stage, the cartridges were transferred to new Eppendorf tubes, and depending on the starting number of cells, 20–35 µL of nuclease-free water was put onto the membrane surface and incubated for three minutes at room temperature. Total RNA isolation was completed with centrifugation at 12,000× g for one minute. The concentrations of RNA samples obtained were determined with NanoDrop Spectrophotometer (Thermo Fisher Scientific, Paisley, UK), and the samples were stored at −80 ◦C in the presence of RNase inhibitors or used for further experiments. Following total RNA isolation, cDNA synthesis was performed using modified MMLV- derived reversible transcriptase using the iScript cDNA Synthesis Kit (BioRad, #1708891, Hercules, CA, USA). For this purpose, a maximum of one µg total RNA sample was diluted to a maximum volume of 15 µL. The RNA sample was denatured at 70 ◦C for five minutes and centrifuged briefly. After the addition of the 5X reaction buffer and iScript reversible transcriptase, the mix was ready for the cycling. Prepared cDNA samples were diluted with nuclease-free water to the desired concentration immediately before use in qPCR and/or stored at −20 ◦C for a maximum of one month. PrimerQuest (Integrated DNA Technologies, IDT, Coralville, IA, USA), a free online software, was used for the qPCR primer design. The mRNA sequences of the target genes were obtained from the NCBI Gene (https://www.ncbi.nlm.nih.gov/gene/, accessed on 20 January 2021) database, the exon regions of the respective genes were determined, and the primers were designed to be at the exon–exon boundary (if possible). Potential primer pairs were evaluated for GC content, melting temperatures (Tm), and the hairpin formation and appropriate primers were determined. The NCBI BLAST database (https: //blast.ncbi.nlm.nih.gov/, accessed on 20 January 2021) was used to check the specificity of the designed primers. The designed primers are listed in Table1. J. Pers. Med. 2021, 11, 125 6 of 28

Table 1. qPCR primers used.

Gene Site Sequence (50-30) Frw CAT CTT CCA GGA GCG AGA TCC GAPDH Rev AAA TGA GCC CCA GCC TTC TCC Frw ACG AAA CTA CCT TCA ACT CC ACTB Rev GAT CTT GAT CTT CAT TGT GCT GG Frw GCT TCC TAC GCA TAC ATT GAC C ELK1 Rev ACT GGA TGG AAA CTG GAA GG Frw GGG AAA TGG GAG GGG TGC AAA AGA GG SOX2 Rev TTG CGT GAG TGT GGA TGG GAT TGG TG Frw AAG GAT GTG GTC CGA GTG TGG POU5F1 Rev CCT GAG AAA GGA GAC CCA GCA G Frw TTC AGA GAC AGA AAT ACC TCA GCC NANOG Rev CCT TCT GCG TCA CAC CAT TGC Frw GACAAAGCTACCAGGGAGTC WNT3A Rev CTGCTGCAGCCACAGAT Frw ACATACTAGAGTTGGCTGCATATT IRAK3 Rev TGTCACCTACACACTGCAATC Frw CAACCGCCTCTTCCAGTATG MEF2B Rev TCAGCGTCTCGAGGATGT Frw TGAGCGTGAAATCACCAGTC TCF7L1 Rev TGGCCCTCATCTCCTTCATA Frw CATGATGAACAAGCAGTTCCG RHO Rev AGAGTCCTAGGCAGGTCTTAG Frw CGGGATCGAGCTGAGAATAG HES7 Rev GTTCCGGAGGTTCTGGTC Frw GCTGGAAGAGTTGGAGAAAGT NOTO Rev ACTCTCACCTGGTTCTCTGTA Frw CAGCAAGAAACGCGAACTG SIX3 Rev GTGCTGGAGCCTGTTCTT Frw ACCTGCATCTTGGTCCTACTA CREB3 Rev GGACAACACTCCATGCTCAG Frw ATCCCAGCATGATGGAAGTATAA CREM Rev ATTGCTGCTACCTGAGCTAAA Frw GCTCTGGACAAGTTAAATCCATAC LIFR Rev CCCTTTGAAGGACTGGCT Frw AAGTTAAGCGCTGGGATATGA FRZB Rev GGGATTTAGTTGCGTGCTTG Frw AGCTCTCTGGGATCAATGCTGTGT GLUT3 Rev ATGGTGGCATAGATGGGCTCTTGA Frw GATGTGAAGCCACCAGTCTTAG RXRB Rev GTAGTGTTTGCCTGAGCTTCT J. Pers. Med. 2021, 11, 125 7 of 28

Table 1. Cont.

Gene Site Sequence (50-30) Frw TACATCCAGAGTCTGCTGAAAC NODAL Rev CTAGGAGCACTCTGCCATTATC Frw GTGAATGGGCGGAGTTATGA PAX6 Rev ATGAGTCCTGTTGAAGTGGTG Frw CCGAGGAGAACCCAATGTTT GSK3B Rev GCCAGCAGACCATACATCTATAC Frw CAAAGGCATCGTCACCAAAC FGF11 Rev GATCAGGTTGAAGTGGGTGAA Frw GTGCAGGAAACCGAGTAGAA FRIT1 Rev GCGCCTTTAGAGTGAGTGAA Frw CTCGGAAGGTCCCAGGT GLI4 Rev CCCGGTGATGAGAGACTGA Frw GTAAACTCCACCAGTCCTACTTT Rev TCTGTCCTTAACAGCTCAACTC Frw GAGGATATCGATGAGTGCAGAAG NOTCH4 Rev TTCAAAGCCTGGGAGACAC Frw GAGCGACAAGCCCTATCTTT ZIC1 Rev GGATTCGTGGACCTTCATGT Frw TCAGCTCATGACTCACCCA ARC Rev CTTGAGACCTGTTGTCACTCTC Frw GGACTCAAAGAAGAGAAGCTCAA ALS2 Rev TGGCAATCTCTCTGGTGTTATG Frw CTTCATGGTGTGGGCTCA SOX10 Rev CGTTCAGCAGCCTCCAG Frw CCTACCGTGTGCTGCAA SMAD6 Rev GGAATCGGACAGATCCAGTG Frw CATAGCCTTCAAGAGGTGGAAC NGFR Rev CACTGTCGCTGTGGAGTTT Frw AGGAGCTTCTCAGCGTAATTC MAPK6 Rev CCAGGAAATCCAGTGCTTCT Frw CATAGCCTTCAAGAGGTGGAAC NGFR Rev CACTGTCGCTGTGGAGTTT Frw GCGTCTTCCTCATGGTTGGAG CD133 Rev CTTGCTCGTGTAAGGTTCACAG

All the qPCR experiments were performed using SSOAdvanced Universal SYBR Green Supermix (Biorad, #1725274, Hercules, CA, USA) and Applied Bioscience StepOne Plus Real-Time PCR System (Thermo Fisher Scientific, Waltham, MA, USA); essentially, 1–10 ng cDNA was used as template; primers were used at 300 nM each, and the reaction was carried out at 60 ◦C for 40 cycles. The differences between the expression of target genes were normalized by the expressions of β-actin and gapdh genes. Each setup was prepared in triplicate and analyzed by the ∆∆CT method as described previously [20]. The fold changes in target gene expressions were calculated based on the mean of the reference J. Pers. Med. 2021, 11, 125 8 of 28

and logarithm-transformed. All qPCR experiments were repeated at least 3 times, unless otherwise noted. The mean and standard deviation values were calculated for each group, and the differences in the gene expression levels were determined considering the control group. For statistical analysis, depending on the context, one-way ANOVA with Tukey post hoc test or Student’s t-test depending on the context with Prism 5 GraphPad software was used. p-value under 0.05 was considered statistically significant. Total RNA was extracted using a Norgen Total RNA isolation kit and quantified using the NanoDrop Spectrophotometer ND-1000. Complementary DNA was synthesized from 0.5–1 µg RNA by using qScript cDNA Super Mix (Quanta Biosciences, Beverly, MA, USA) and a C1000 Thermo Cycler (Bio-Rad, Hercules, CA, USA) with the following cycle param- eters: 4 min at 25 ◦C, 30 min at 42 ◦C, 5 min at 85 ◦C, hold at 4 ◦C. qRT-PCR was performed by using Perfecta SybrGreen (Quanta Biosciences, Waltham, MA, USA) and an Opticon Chroma4 instrument (Bio-Rad, Hercules, CA, USA). Gene expression was quantified by using Opticon software, and expression levels were normalized to 28srRNA expression. For statistical analysis, multiple Student’s t-tests with Prism 5 GraphPad software were used. p-value under 0.05 was considered statistically significant.

2.7. Microarray and Data Analysis For microarray analysis, SH-SY5Y cells were transfected with Elk-1-VP16 expression plasmid or empty pCDNA3 plasmid as described above, and 48 h after transfection, RNA samples were isolated using Ambion Tri-pure RNA isolation kit, checked for quality, con- verted to cDNA, and confirmed for Elk-1 expression as described above. Thereafter, RNA was converted to cDNA using the Superscript Double-Stranded cDNA Synthesis (Invit- rogen, Carlsbad, CA, USA) Kit and labeled with NimbleGen One Color DNA Labeling (NimbleGen, Roche, Madison, WI, USA). The labeled cDNA was hybridized to NimbleGen Human Gene Expression Array 12x135K (NimbleGen, Roche, Wisconsin, USA), which covers 45.033 genes with 3 probes per gene, containing 12 arrays per slide. After hybridiza- tion, slides were scanned using Genepix 4000B scanner and analyzed with NimbleScan 2.5 software using three arrays from the pCDNA3-transfected cell as reference samples. The expression datasets were normalized using the Robust Multi-Array Average expression measure [21], and differentially expressed genes (DEGs) and their fold-changes were iden- tified from the normalized expression values using two-tailed Student’s t-test assuming equal variances and Benjamini-Hochberg’s method as the multiple testing option to control the false discovery rate. An adjusted p-value threshold of 0.15 was used to determine the statistical significance of differential expression. The dataset is accessible from EBI ArrayExpress, with the accession number of E-MTAB-9938. Gene IDs were converted to official gene symbol, and gene set enrichment analyses of DEGs were performed through ConsensusPathDb (r.32) [22] using KEGG [23], Reac- tome [24], and Biocarta [25] as the data source for molecular pathways, and Biological Process annotations [26] as the data source for biological processes. Whole- genome annotation for the was used as the background reference set. p-values were determined through a modified Fisher exact test and adjusted via Benjamini- Hochberg’s method. A threshold of adjusted p-value < 0.05 was used to determine the statistical significance of the enrichment results. Besides, to characterize the molecular functions of each gene product, and their association with diseases, we manually searched GeneCards Human Gene Database [27].

2.8. Promoter Clonings and Site-Directed Mutagenesis To identify the putative Elk-1 transcription factor binding sites in selected stem- ness gene promoters (SOX2, NANOG, POU5F1), the Cold Spring Harbor Laboratory— Transcriptional Regulatory Element Database (TRED), Swiss Institute of Bioinformatics— The Eukaryotic Promoter Database (EPD), and Alggen-Promo algorithmic analysis pro- gram were used. The promoter sequences that correspond to the genes of interest were retrieved from either the Transcriptional Regulatory Element Database (TRED) (http: J. Pers. Med. 2021, 11, 125 9 of 28

//rulai.cshl.edu/cgi-bin/TRED/tred.cgi?process=home, accessed on 20 January 2021), or the Eukaryotic Promoter Database (EPD) (http://epd.vital-it.ch/, accessed on 20 Jan- uary 2021). The obtained promoter sequences were analyzed with Promo 3.0 (http: //alggen.lsi.upc.es/cgi-bin/promo_v3/promo/promoinit.cgi?dirDB=TF_8.3, accessed on 20 January 2021). The promoter binding regions for transcription factors can be analyzed by the Promo 3.0 tool, and the results are displayed as “dissimilarity rate”. The dissimilarity matrix expresses the similarity pair to pair between Elk-1 DNA binding sequence and the putative sequences at analyzed genes. From this point of view, the smaller dissimilarity rates are the indicators of a higher possibility for the interaction between Elk-1 and the promoter of interest. The binding ability of Elk-1 to the predicted sites on the promoters could be confirmed by luciferase and chromatin immunoprecipitation assays, thereby verifying the microarray results (Table2).

Table 2. Number of ets motifs predicted on selected promoters and their dissimilarity score (DS) range; DS of 0% means perfect match to consensus; TRED, Transcriptional Regulatory Element Database; EPD, Eukaryotic Promoter Database. *

Number of Predicted ets Binding Motifs with Different Dissimilarity Scores (DS) in Promo 3.0 DS: 0–1 Percent DS: 1–5 Percent DS: 5–10 Percent SRF - 2 1 MCL1 2 - 3 LIF - 2 1 SOX2 (TRED) - 1 1 NANOG (TRED) - 1 - POU5F1 (TRED) - 1 2 SOX2 (EPD) 1 2 3 NANOG (EPD) - 1 1 POU5F1 (EPD) 1 2 1 * For dissimilarity scores of individual ets motifs, see Supplementary Table S3 for URL of databases, please refer to text.

Cloning primers for human SOX2, NANOG, and POU5F1 promoters were designed and analyzed with NetPrimer (http://www.premierbiosoft.com/netprimer/, accessed on 20 January 2021) and PrimerBlast (http://www.ncbi.nlm.nih.gov/tools/primer-blast/, accessed on 20 January 2021) softwares. The designed cloning primers are listed in Table3 . Gradient PCR with five different annealing temperatures was performed to detect the optimum annealing temperature of the primers. The PCR reactions were prepared with i-Taq DNA Polymerase (Intron, #25024, Seoul, Korea) kit using the genomic DNA isolated from the SH-SY5Y cell line as a template. After optimization, the preparation of the insert was carried out using Pfu DNA Polymerase (Thermo Scientific, #EP0571, Waltham, MA, USA) suitable annealing temperatures as indicated in text for 30 cycles. Following amplification, PCR products were purified by PureLink PCR Purification Kit (Invitrogen, # K3100-01) and cloned into pGL3luciferase reporter plasmid. Intentional deletion mutations were made on cloned promoter sequences with site- directed mutagenesis (SDM). The promoter sequences were analyzed with Promo 3.0, as stated previously. Potential Elk-1 binding sites on stemness promoters were chosen according to the dissimilarity rate Promo 3.0. Accordingly, motif on NANOG promoter, ets1 and motifs on SOX2 promoter, and ets1, ets2, and ets3 motifs on POU5F1 promoter were deleted in corresponding pGL3 luciferase reporter constructs. SDM primers were designed using the NEB Base Changer (http://nebasechanger.neb.com/, accessed on 20 January 2021) website, and Q5® Site-Directed Mutagenesis Kit Protocol (NEB, #E0554) was followed for the mutations. The primer pairs designed flanking the region to be deleted J. Pers. Med. 2021, 11, 125 10 of 28

and eventually forming deletion mutants from the cloned promoter sequences are given in Table4. The mutagenesis was carried out according to the manufacturer’s instructions.

Table 3. Cloning primers for chosen Homo sapiens stemness gene promoters.

Forward Primer Reverse Primer Product Promoter RE Site RE Site (50-30) (50-30) (bp) AGACggtaccAGGGCTG CTGTagatctAGCCATTTAA POU5F1 KpnI BglII 993 TTGGCTTTGGACA GAATTCCAGAGTAGG CTGTggtaccGGGGAGTG CTGTagatctCACTAGACTG SOX2 KpnI BglII 993 ATTATGGGAAGAA TCTTCATTCAACCGTAGC CTGTggtaccTTTCTGCC CTGTagatctAGGTGAAGA NANOG KpnI BglII 988 TAAACTAGCCA TTCTTTACAGTCG

Table 4. Primers used for site-directed mutagenesis of cloned promoters.

Primer Site Oligo Length Tm (◦C) Ta (◦C) Fwd TACTAACATGAGTGTGGATC 20 59 NANOG-ets∆ 58 Rev AGGAGGAAAAAATTTAAGAGG 21 57 Fwd CCTTTCCCCCTGTCTCTG 18 64 POU5F1- ets∆1 65 Rev CAGGGAAAGGGACCGAGG 18 68 Fwd GAATTGGGAACACAAAGG 18 57 POU5F1- ets∆2 57 Rev TGAATGAAGAACTTAATCCC 20 56 Fwd GTGAAGTTCAATGATGCTCTTG 22 61 POU5F1- ets∆3 62 Rev AACCAGTTGCCCCAAACT 18 64 Fwd TTGAAATCACCCTCCCCC 18 64 SOX2- ets∆1 65 Rev ATCCCACGGCACTGTATG 18 65 Fwd GTGTCTTTCCCCAGCCCC 18 69 SOX2- ets∆2 68 Rev GGCGCTCAAAAGTGCAGG 18 67

2.9. Luciferase Reporter Assay For each cell line, the necessary optimization experiments were performed, and cell numbers and DNA: PEI ratios were determined for co-transfections. For 24-well cell culture plates for luciferase analysis, for SK-N-BE (2), T98G, and A172 cells 80 × 104 cells/well, and for SH-SY5Y and U87-MG cells, 60 × 104 cells/well were seeded with triplicates for each transfection group. The following day, SOX2-luc, NANOG-luc, POU5F1-luc, or one of the deletion mutant of these plasmids, one of the Elk-1 series plasmids (pCDNA3.1, Elk1-VP16, Elk1-EN, siElk1, or scrRNA), Renilla-luc plasmid (pRL-TK (Promega, #E2241, Madison, WI, USA)) and the proper ratio of PEI mixture was prepared. After transfection, the cells were incubated for 42 h in a normoxic medium and subjected to one percent hypoxia for the last six hours for the normoxia–hypoxia experiments. At the end of hypoxia treatment, luciferase analysis was performed with Thermo Luminoskan Ascent device by using Dual-Glo Luciferase kit (Promega, Wisconsin, USA) with some modifications. For luciferase analysis of monolayer cell lines, 48 h of incubation was necessary before performing luciferase analysis. On the day of the luciferase assay, the medium on the cells was aspirated, and the wells were washed with PBS. Cells were lysed with 100 µL of 5X Passive Lysis Buffer (PLB) (Promega, #E1941, Wisconsin, USA) diluted to 1X. Seventy-five microliters of the cells were transferred to luminometer compatible white-bottomed 96-well plates. To measure the Firefly luciferase activity, 75 µL of Dual-Glo® Luciferase Reagent was added onto the J. Pers. Med. 2021, 11, 125 11 of 28

lysed cells. For at least 15 min, the plates were incubated at room temperature, and the luminescence for Firefly luciferase activity was measured. To measure the Renilla luciferase activity, 75 µL of Dual-Glo® Stop&Glo Luciferase Reagent was added to the wells. They were incubated at room temperature for the equal time that was done for Firefly luciferase, and the luminescence for Renilla luciferase activity was measured. Firefly/Renilla ratios were calculated, normalizations were done, and the results were graphed as relative luciferase activity. For statistical analysis, one-way ANOVA with Tukey post hoc test or Student’s t-test depending on the context with Prism 5 GraphPad software was used. p-value under 0.05 was considered significant.

2.10. Chromatin Immunoprecipitation (ChIP) Assay In this assay, proteins and interacting DNA are crosslinked with formaldehyde; the chromatin is sheared with either sonication mechanically or micrococcal nuclease enzy- matically. The nucleoprotein complex is enriched by immunoprecipitation, and through the reversal of the crosslinking, DNA and the interacting protein are separated. In the end, the interacting DNA fragment is purified and quantified with ChIP-qPCR. To de- termine the promoter fragment to be amplified in ChIP PCR, Promo3.0 analysis used for predicting ets motifs and their dissimilarity scores was used (Supplementary Table S3). The amplicon size was arranged between 75–150 bp; CpG islands were checked for the potential binding sites of Elk-1, and the position of Elk-1 to those sequences was con- sidered for primer design. The UCSC in silico PCR tool was used to verify the ampli- con (https://genome.ucsc.edu/cgi-bin/hgPcr, accessed on 20 January 2021); primers used for ChIP PCR are listed in Table5.

Table 5. The list of primers used in chromatin immunoprecipitation (ChIP) assay.

Name Primer Sequence PCR Product Size Frw GCCGCCCTAAAACCGTGATA ChIP_MCL1 99 Rev CGCCTGGCTGAGAAAACTG Frw TGACAGCAACGAGTTCGGTA ChIP_SRF 130 Rev CCCCCATATAAAGAGATACAATGTT Frw TGGGAGGGAGTTTGTGACT ChIP_SOX2_ETS1 97 Rev AAAGTGCAGGCGATGGG Frw GTGGGATGCCAGGAAGTT ChIP_SOX2_ETS2 102 Rev GTCGTGCGGCTTTCAAATG Frw AGACAGTCTAGTGGGAGATGTG ChIP_SOX2_ETS3 138 Rev CGGACCATAAGGCAGACTCTA Frw CTTATGGTCCGAGCAGGATTT ChIP_SOX2_ETS4 103 Rev TCCCGACTAGAAGTTAGGAGAC Frw CGCACCTTAGCTGCTTCC ChIP_SOX2_ETS5 143 Rev GTCACACCACACGCCTTT Frw CTGGAGGTCCTATTTCTCTAACATC ChIP_NANOG_ETS1 155 Rev ATGCTTCAAAGCAAGGCAAG Frw GCAGAGGAGAATGAGTCAAAGA ChIP_NANOG_ETS2 131 Rev CCCAAACCCAACATTCAAGAAA Frw CTTAGTCCAGCCTGTTCCAAA ChIP_NANOG_ETS3 136 Rev AGTGAAAGACCAAAGGGAAGG Frw CTTCACTGCACTGTACTCCTC ChIP_POU5F1_ETS1 101 Rev CACCTCAGTTTGAATGCATGG Frw GGAGTTTGTGCCAGGGTT ChIP_POU5F1_ETS2 105 Rev CCCTCCAACCAGTTGCC Frw GTTGGAGGGAAGGTGAAGTT ChIP_POU5F1_ETS3 93 Rev TACTGTGTCCCAAGCTTCTTTAT J. Pers. Med. 2021, 11, 125 12 of 28

Table 5. Cont.

Name Primer Sequence PCR Product Size Frw GCCGCCCTAAAACCGTGATA ChIP_MCL1 99 Rev CGCCTGGCTGAGAAAACTG Frw TGACAGCAACGAGTTCGGTA ChIP_SRF 130 Rev CCCCCATATAAAGAGATACAATGTT Frw TGGGAGGGAGTTTGTGACT ChIP_SOX2_ETS1 97 Rev AAAGTGCAGGCGATGGG Frw GTGGGATGCCAGGAAGTT ChIP_SOX2_ETS2 102 Rev GTCGTGCGGCTTTCAAATG Frw AGACAGTCTAGTGGGAGATGTG ChIP_SOX2_ETS3 138 Rev CGGACCATAAGGCAGACTCTA Frw CTTATGGTCCGAGCAGGATTT ChIP_SOX2_ETS4 103 Rev TCCCGACTAGAAGTTAGGAGAC Frw CGCACCTTAGCTGCTTCC ChIP_SOX2_ETS5 143 Rev GTCACACCACACGCCTTT Frw CTGGAGGTCCTATTTCTCTAACATC ChIP_NANOG_ETS1 155 Rev ATGCTTCAAAGCAAGGCAAG Frw GCAGAGGAGAATGAGTCAAAGA ChIP_NANOG_ETS2 131 Rev CCCAAACCCAACATTCAAGAAA Frw CTTAGTCCAGCCTGTTCCAAA ChIP_NANOG_ETS3 136 Rev AGTGAAAGACCAAAGGGAAGG Frw CTTCACTGCACTGTACTCCTC ChIP_POU5F1_ETS1 101 Rev CACCTCAGTTTGAATGCATGG Frw GGAGTTTGTGCCAGGGTT ChIP_POU5F1_ETS2 105 Rev CCCTCCAACCAGTTGCC Frw GTTGGAGGGAAGGTGAAGTT ChIP_POU5F1_ETS3 93 Rev TACTGTGTCCCAAGCTTCTTTAT

Essentially, cells were seeded in three separate 150 mm cell culture dishes of 2 × 106 cells/dish per experimental group on day zero. On day 1, cells were transfected with either an empty pCDNA3.1 plasmid or an expression plasmid for Elk1-VP16 plasmids and incubated ◦ 48 h at 37 C, 5% CO2. Cells were then treated with 1% formaldehyde at room temperature for 20 min; glycine was then added to the dishes to a final concentration of 0.125 M and incubated for 5 min at room temperature. The dishes were washed three times with cold PBS on ice and then centrifuged at 400× g for five minutes at 4 ◦C with 1× protease inhibitor cocktail (PIC) (Roche, 4693159001). The supernatant was aspirated, and lysis buffer was added onto the cells with a volume of at least 10 times the pellet obtained. The suspension was incubated on ice for 10 min and passed through an insulin needle 20 times. One volume of the sample was mixed with an equal volume of 0.4 percent Trypan Blue Dye, and the cell nuclei were checked under the microscope. The volume of the sonication buffer to be used to dissolve the pellet was adjusted to 2–3 × 106 nuclei/mL and sonicated in the Biorupter UCD-200 Sonicator (Diagenode, Denville, NJ, USA). Following the sonication, cell lysates were centrifuged at 22,000× g for 20 min at 4 ◦C to remove insoluble materials. The supernatant was then diluted five-fold with dilution buffer and pre-cleared for 4 h with slow rotation with protein A/G mixture beads. After incubation, the samples were precipitated at 150× g for 5 min at 4 ◦C, and 10% of the total supernatant was removed as total input control and kept in −20 ◦C. The rest of the supernatant was divided into two fractions of the negative control (IgG-mock) and immunoprecipitation (IP) per group. Sixty microliters of ANTI-FLAG® M2 Affinity Gel (Sigma Aldrich, #A2220, Taufkirchen, Germany) resin per group were washed and equilibrated with five volumes of dilution buffer and centrifuged three times at 400× g for one minute each at 4 ◦C. The negative control and J. Pers. Med. 2021, 11, 125 13 of 28

IP fractions separated from the dilution in the previous step were mixed with Protein G-Plus agarose beads and anti-Flag M2 resin, respectively. The tubes were incubated at 4 ◦C overnight with slow rotation. The following day, the mix was centrifuged at 4 ◦C and 600× g for five minutes, and the pellet was collected. The beads were washed with one mL of low salt, high salt, LiCl, and TE buffers at 4 ◦C with rotation, respectively. Following each of the washing steps, the beads were centrifuged at 4 ◦C and 600× g for five minutes. At the elution step, the inputs that were collected and frozen a day before were thawed and added as the third fraction of each group. After the last wash, 250 µL fresh elution buffer, pre-heated at 65 ◦C, was added onto the beads, and they were incubated on a shaker for 15 min. The tubes were vortexed with five-minute intervals and then centrifuged at 4 ◦C and 18,000× g for five minutes. The supernatant was collected for each fraction of each group, and the elution step was repeated with another 250 µL elution buffer. After elution of the crosslinked DNA–protein complex, 10 µL of RnaseA (10 mg/mL) (Intron, #BR003) and 25 µL of 5 M NaCl was added onto the elutes and incubated for at least five hours or overnight at 65 ◦C. The following day, 10 µL of 0.5 M EDTA, 20 µL 1 M Tris–HCl (pH 6.5), and two µL Proteinase K (20 mg/mL) (Invitrogen, #25530049, Carlsbad, CA, USA) mix were added and incubated again at 65 ◦C for two more hours. Using MEGAquick-spin™ Plus Total Fragment DNA Purification Kit (Intron Bio, #17290, Sungnam, Korea), the DNA was cleaned up. The resulting fractions were used for qPCR analysis. SSOAdvanced Universal SYBR Green Supermix (Bio-Rad, #1725274, Hercules, CA, USA) and Applied Biosciences StepOne Plus Real-Time System were used for qPCR analysis with DNA isolated from ChIP. Ten microliters of PCR reaction were prepared by mixing 2X SSO Advanced Universal SYBR Green Supermix, 300 nM forward and reverse primers each, and 1 µL template. In the analysis phase, qPCR signals obtained from the ChIP samples were normalized by the signals obtained from the input, and the mock samples and the results are presented as fold change. For statistical analysis, one-way ANOVA with Tukey post hoc test or Student’s t-test depending on the context with Prism 5 GraphPad software was used. p value under 0.05 was considered significant.

3. Results 3.1. Microarray Analyses Reveal Novel Targets in Elk-1-VP16 Overexpressing SH-SY5Y Cells Elk-1 is a ubiquitous transcription factor, yet it has been implicated in different bio- logical processes in the nervous system. In order to identify novel target genes of Elk-1 with respect to survival in neurons, we have overexpressed Elk-1-VP16 constitutively active fusion protein in SH-SY5Y neuroblastoma cells. The comparative analysis of the transcriptome profiles indicated 11,018 differentially expressed genes (DEGs), of which 4212 were downregulated and 6806 were upregulated, when SH-SY5Y neuroblastoma cells were transfected with Elk1-VP16. The gene set enrichment analysis (GSEA) of these genes up- or downregulated by exogenous Elk-1-VP16 presented overrepresentation of quite a high number of biological processes such as anatomical structure development, cell proliferation, single-organism developmental process, developmental growth, and organ and tissue development, including forebrain and midbrain development (Supplement Tables S1 and S2). When a subset of these genes was analyzed further, stemness genes such as POU5F1, SOX2, and NANOG, as well as growth factors and receptors or transcription factors including FGFR1, WNT16, WNT 3, PDGFA, PAX6, PAX7, HIF3A, NOTO, among many others were found to be upregulated, whereas genes such as EGLN2, FEV, JUNB, and GLI4 were found to be downregulated upon overexpression of Elk-1-VP16 (Figure1A,B). Prediction of putative Elk-1 binding sites (i.e., ets motifs) on the promoters of these genes was assessed via Alggen PROMO 3.0 online software [28]. Among the genes of interest for which human promoter sequences were available, the analysis was performed for human ELK-1 (TRANSFAC database accession no. T00250) binding, thereby limiting the number of promoters investigated, and out of these, promoters with at least one motif are listed (Supplement Table S3). Among the selected subset of genes, SOX2 promoter was found to contain one ets motif with a dissimilarity score of 2.16, NANOG was found to J. Pers. Med. 2021, 11, 125 13 of 27

and incubated again at 65 °C for two more hours. Using MEGAquick-spin™ Plus Total Frag- ment DNA Purification Kit (Intron Bio, #17290, Sungnam, Korea), the DNA was cleaned up. The resulting fractions were used for qPCR analysis. SSOAdvanced Universal SYBR Green Supermix (Bio-Rad, #1725274, California, USA) and Applied Biosciences StepOne Plus Real-Time System were used for qPCR analysis with DNA isolated from ChIP. Ten microliters of PCR reaction were prepared by mixing 2X SSO Advanced Universal SYBR Green Supermix, 300 nM forward and reverse primers each, and 1 µL template. In the analysis phase, qPCR signals obtained from the ChIP sam- ples were normalized by the signals obtained from the input, and the mock samples and the results are presented as fold change. For statistical analysis, one-way ANOVA with Tukey post hoc test or Student’s t-test depending on the context with Prism 5 GraphPad software was used. p value under 0.05 was considered significant.

3. Results 3.1. Microarray Analyses Reveal Novel Targets in Elk-1-VP16 Overexpressing SH-SY5Y Cells Elk-1 is a ubiquitous transcription factor, yet it has been implicated in different biolog- ical processes in the nervous system. In order to identify novel target genes of Elk-1 with respect to survival in neurons, we have overexpressed Elk-1-VP16 constitutively active fu- J. Pers. Med. 2021, 11, 125 sion protein in SH-SY5Y neuroblastoma cells. The comparative analysis of the transcriptome14 of 28 profiles indicated 11,018 differentially expressed genes (DEGs), of which 4212 were down- regulated and 6806 were upregulated, when SH-SY5Y neuroblastoma cells were transfected with Elk1-VP16. The gene set enrichment analysis (GSEA) of these genes up- or downregu- containlated oneby exogenous ets motif Elk-1-VP16 with a dissimilarity presented overrepr score ofesentation 2.3, and POU5F1 of quite acontained high number one of etsbi- motif withological a dissimilarity processes scoresuch as of anatomical 3.12, among struct otherure potentialdevelopment, ets bindingcell proliferation, sites, indicating single-or- a high probabilityganism developmental of binding (Supplement process, developmenta Table S3).l Other growth, promoters and organ of and the tissue microarray-determined development, setincluding of putative forebrain Elk-1 targetand midbrain genes, developmen whose promoterst (Supplement contained Tables low S1 dissimilarityand S2). When score a ets subset of these genes was analyzed further, stemness genes such as POU5F1, SOX2, and motifs, included transcription factors such as RXRB, TCF7L1, MEF2B, PAX6, SOX10, CREB3, NANOG, as well as growth factors and receptors or transcription factors including FGFR1, SMAD6, CREM, and HES7 and signal transduction pathway elements such as RHO, IRAK3, WNT16, WNT 3, PDGFA, PAX6, PAX7, HIF3A, NOTO, among many others were found to WNT3Abe upregulated,, LIFR, FRZB whereas, NGFR genes, MAPK6 such as, NOTCH4 EGLN2, ,FEV,FGF11 JUNB,, and andNODAL GLI4 were, among found many to be others (Supplementdownregulated Table upon S3). overexpression of Elk-1-VP16 (Figure 1A,B).

FigureFigure 1. (A) 1. Heatmap (A) Heatmap of a subsetof a subset of genes of genes regulated regulated by Elk-1-VP16by Elk-1-VP16 showing showing increased increased (green) (green) or or decreased decreased (red)(red) expressionexpres- in Elk1-VP16sion in overexpressing Elk1-VP16 overexpressing SH-SY5Y neuroblastoma SH-SY5Y neuroblastom cells; 557A10,a cells; 557A11,557A10, 557A12557A11, correspond557A12 correspond to SH-SY5Y to SH-SY5Y cells transfected cells transfected with Elk-1-VP16 expression plasmid, 557A01,557A02,557A03 control SH-SY5Y cells transfected with empty with Elk-1-VP16 expression plasmid, 557A01,557A02,557A03 control SH-SY5Y cells transfected with empty plasmid; color key shows up- and downregulation levels. (B) Schematic representation of the relation between selected genes in pluripotency and early embryonic development pathways that were found to be regulated by Elk-1-VP16 in microarray analysis.

3.2. Regulation of Nervous System Development Related Genes by Elk-1 To validate regulation of selected candidate genes identified through microarray experiments by Elk-1 transcription factor, we have either overexpressed Elk-1-VP16 consti- tutively active fusion protein or knocked down endogenous Elk-1 expression in SH-SY5Y and SK-N-BE (2) neuroblastoma cell lines and A172 and T98G GBM cell lines (Figure2). qPCR results in SH-SY5Y cells were parallel to those obtained from the microarray analysis, especially in the genes related to pluripotency such as SOX2, NANOG, POU5F1, RXRB, GLUT3, TCF7L1, NODAL, and CREB3 (Figure2A,B). SOX2 was upregulated in SH- SY5Y overexpressing Elk-1-VP16 protein, similar to microarray, but not in other cell types, while it was repressed when Elk-1 was knocked down (siElk-1) in all cell types (Figure2). Similarly, NANOG and POU5F1 was upregulated in SH-SY5Y cell overexpressing Elk-1-VP16, but downregulated in cells transfected with siElk-1 plasmid (Figure2A,B), whereas both genes were repressed in SK-N-BE (2) cells overexpressing Elk-1-VP16 and upregulated in siElk-1 knockdown (Figure2C,D; Table6), indicating a cell context-dependent regulation. TCF7L1 and NODAL expression increased in Elk-1-VP16 overexpressing SH-SY5Y and SK-N-BE (2) but decreased in siElk-1 silencing; BRACHYURY (T) expression was upregulated in Elk-1-VP16 overexpressing but decreased in siElk-1 silenced SK-N-BE (2) cells (Figure2; Table6). GLUT3 expression was upregulated in all cell types overexpressing Elk-1-VP16, and decreased in all cells with siElk-1 silencing, paralleling the microarray results (Figure2, Table6). The expression of ARC and CREB3 increased in A172 and T98G cells overexpressing Elk-1-VP16 but decreased in siElk-1 knockdown cells (Figure2E–H; Table6) . GLI4 and ALS genes increased in A172 cells overexpressing Elk-1-VP16 and decreased with siElk-1 silencing (Figure2E,F). J. Pers. Med. 2021, 11, 125 15 of 27 J. Pers. Med. 2021, 11, 125 15 of 28

Figure 2. qPCR expression profiles of selected genes in different cell lines upon overexpression of Elk-1-VP16 (A,C,E,G) or Figure 2. qPCR expression profiles of selected genes in different cell lines upon overexpression of Elk-1-VP16 (A,C,E,G) knockdown of endogenous Elk-1 (B,D,F,H). Expression profiles after (A). over-expression with Elk1-VP16 and (B). after or knockdown of endogenous Elk-1 (B,D,F,H). Expression profiles after (A). over-expression with Elk1-VP16 and (B). after knock-downknock-down withwith siElk1 siElk1 in in SH-SY5Y SH-SY5Y neuroblastoma neuroblastoma cell cell line; line; expression expression profiles profiles after after (C). over-expression(C). over-expression with Elk1-VP16with Elk1- andVP16 (D and). after (D). knock-down after knock-down with siElk1 with in siElk1 SK-N-BE(2) in SK-N-BE(2) neuroblastoma neuroblastoma cell line; expressioncell line; expression profiles after profiles (E). over-expression after (E). over- withexpression Elk1-VP16 with and Elk1-VP16 (F). after and knock-down (F). after withknock-down siElk1 in with A172 siElk1 GBM cellin A172 line; expressionGBM cell line; profiles expression after (G ).profiles over-expression after (G). withover-expression Elk1-VP16 and with (H Elk1-VP16). after knock-down and (H). after with siElk1knock-down in T98G with GBM siElk1 cell line.in T98G Unpaired GBM t-test;cell line. **** Unpairedp < 0.0001, t-test; *** p <**** 0.001, p < **0.0001,p < 0.01, *** *p p< <0.001, 0.05. ** p < 0.01, * p < 0.05.

J. Pers. Med. 2021, 11, 125 16 of 28

Table 6. Summary of qPCR and microarray comparisons of selected potential Elk-1 target genes after Elk1-VP16 over- expression or siElk-1 silencing in neuroblastoma and glioblastoma cell lines.

Elk-1-VP16 Overexpression siElk-1 Silencing Gene ID SH-SY5Y SK-N-BE (2) T98G A172 Microarray Data SH-SY5Y SK-N-BE (2) T98G A172 ALS2 N/A* N/A −0.5 0.60 −6.06 N/A N/A −0.64 −1.44 ARC N/A N/A 0.44 1.67 −6.87 N/A N/A −2.17 −0.06 BRACHYURY N/A 0.32 N/A N/A 1.54 N/A −0.25 N/A N/A CREB3 −0.01 N/A 0.83 0.46 −1.92 0.09 N/A −0.74 −0.67 CREM N/A N/A −0.91 1.40 N/A −0.51 N/A −0.43 0.44 ELK-1 1.50 2.04 8.48 8.98 13.11 −0.54 −2.36 −0.60 −1.49 FGF11 −0.23 N/A N/A 0.28 −2.40 0.48 N/A N/A −1.31 FRIT1 0.17 N/A N/A N/A 1.82 −0.34 N/A N/A N/A FRZB 1.21 −0.72 −0.86 N/A 1.91 −2.20 −1.34 −0.18 N/A GLI4 N/A N/A N/A 0.695 −3.81 N/A N/A N/A 0.61 GLUT3 1.62 0.07 0.83 0.92 2.43 −0.67 −0.46 0.54 2.64 GSK3B −0.22 N/A N/A N/A −1.61 −0.12 N/A N/A N/A HES7 N/A N/A 0.68 N/A −1.77 0.12 N/A 0.20 N/A IRAK3 2.78 0.51 −0.46 N/A 1.70 −0.23 −0.81 −0.97 N/A LIFR −0.91 −1.52 −0.84 −0.08 −2.01 −0.53 0.48 −0.12 −0.82 MAPK6 N/A N/A N/A 0.76 1.64 N/A N/A N/A N/A MEF2B 0.09 −0.41 −1.19 N/A −2.74 −0.19 −0.33 −1.57 N/A NANOG 1.96 −0.91 0.58 1.18 2.54 −0.99 4.85 −0.07 3.12 NODAL 2.33 0.66 N/A N/A 1.64 −0.66 −0.26 N/A N/A NOTCH4 N/A N/A N/A 3.56 3.39 N/A N/A N/A 7.16 NOTO −2.94 0.18 −0.34 N/A 2.15 1.05 −0.08 −0.76 N/A PAX6 −1.54 −0.28 N/A N/A 2.61 1.18 −1.62 N/A N/A POU5F1 1.62 −1.20 0.56 0.32 3.68 −2.19 3.31 −0.67 1.06 RHO −3.56 −1.07 −2.26 N/A 2.27 −0.20 −0.30 −0.93 N/A RXRB 1.01 0.66 0.83 0.46 5.95 0.45 0.31 N/A N/A SIX3 −1.94 N/A 0.61 N/A −5.72 1.56 N/A 0.41 N/A SMAD6 N/A N/A N/A −0.29 −3.78 N/A N/A N/A −0.42 SOX2 1.54 0.41 0.37 −0.59 2.75 −0.84 −1.08 −0.56 −0.36 SOX10 N/A N/A N/A 2.94 2.41 N/A N/A N/A 5.83 TCF7L1 0.79 0.56 0.15 N/A 2.34 −0.29 −0.25 0.14 N/A WNT3A −2.11 0.53 0.11 N/A 2.25 0.86 −1.11 0.52 N/A ZIC1 N/A 1.53 N/A N/A 2.26 N/A 1.07 N/A N/A * N/A: the expression level is not available.

The promoters of a subset of genes have been selected for chromatin immunoprecipi- tation to address whether predicted binding sites were indeed binding to Elk-1 (Figure3 ). To that end, we have transfected SK-N-BE (2) neuroblastoma and T98G GBM cell lines with Elk-1-Flag expression vector and pulled down exogenous Elk-1 using Flag-agarose beads. The known targets SRF (p = 0.0451) and MCL1 (p = 0.0102) showed significant binding in SK-N-BE (2) cells, but the binding was not statistically significant in T98G cells (Figure3A). Among the novel promoters identified in this study, GLUT3 promoter showed Elk-1 binding in both cell types, albeit not to the same extent, while (p = 0.0496) only showed significant binding in T98G cells (Figure3B). LIF1, however, did not show significant Elk-1 binding in either cell type. J.J. Pers.Pers. Med.Med. 20212021,, 1111,, 125125 1717 ofof 2827

Figure 3. Chromatin immunoprecipitation assay for the the identification identification of of Elk-1 Elk-1 binding binding sites sites on on the the target gene promoters in in pCMV-transfected pCMV-transfected (pCM (pCMV)V) vs. vs. Elk-1 Elk-1 over-expressing over-expressing cells cells (Elk-1) (Elk-1) in in (A (A). ). SK-N-BE (2) cells and (B). T98G cells. Lysates were immunoprecipitated with either Flag antibody SK-N-BE (2) cells and (B). T98G cells. Lysates were immunoprecipitated with either Flag antibody (Flag IP) for exogenous Elk-1 or IgG (IgG IP) as control. qPCR results were analyzed as explained (Flag IP) for exogenous Elk-1 or IgG (IgG IP) as control. qPCR results were analyzed as explained in in Materials and Methods and reported as average fold change. Materials and Methods and reported as average fold change. 3.3. Regulation of SOX2,SOX2, NANOG,NANOG, and POU5F1 by Elk-1 in CD133+ Cells Since Elk-1 was previously shown to bebe importantimportant inin humanhuman embryonicembryonic stemstem cellscells (hESCs) maintenancemaintenance of self-renewalself-renewal capacitycapacity throughthrough co-occupationco-occupation ofof promoterspromoters withwith ERK2 [[29],29], andand toto regulateregulate the the promoter promoter of of CD133, CD133, a cella cell surface surface protein protein commonly commonly used used as a cancer stem cell marker [15], we addressed whether the cell context-dependent regulation

J. Pers. Med. 2021, 11, 125 18 of 27

J. Pers. Med. 2021, 11, 125 18 of 28 as a cancer stem cell marker [15], we addressed whether the cell context-dependent regu- lation was due to heterogenous nature of some cell lines used in terms of their tu- morsphere forming abilities. SK-N-BE (2) neuroblastoma cells were shown to form was due to heterogenous nature of some cell lines used in terms of their tumorsphere CD133+ tumorspheres, unlike SH-SY5Y cells, hence we have first sorted CD133− and forming abilities. SK-N-BE (2) neuroblastoma cells were shown to form CD133+ tumor- CD133+spheres, SK-N-BE unlike SH-SY5Y (2) cells cells,and showed hence we that have expression first sorted of CD133,− ELKand-1 CD133+, SOX2, NANOG SK-N-BE, and(2) cells POU5F1 andshowed were all that significantly expression more of CD133 in CD133+, ELK -cells1, SOX2 than, NANOGin CD133,− and cellsPOU5F1 (Figurewere 4A). Intriguingly,all significantly ELK-1 more levels in CD133+ increased cells thanin diff inerent CD133 passages− cells (Figure(p1 and4A). p2) Intriguingly, of CD133+ ELK-1sorted cells,levels while increased CD133 in levels different declined passages with (p1 each and passage; p2) of CD133+ NANOG sorted and POU5F1 cells, while levelsCD133 also increasedlevels declined slightly with in eachp2 cells, passage; albeitNANOG not significantlyand POU5F1 (Figurelevels 4B). also Both increased passages slightly (p1 and in p2)p2 cells,of CD133+ albeit notSK-N-BE significantly (2) cells (Figure were 4shownB). Both to passagesbe Nestin+ (p1 (data and p2)not ofshown). CD133+ To SK-N-BE address whether(2) cells werethis coexpression shown to be Nestin+ of ELK-1 (data with not stemness shown). genes To address studied whether is through this coexpressiondirect regu- lation,of ELK-1 wewith have stemness silenced genesendogenous studied Elk-1 is through expression direct in regulation, CD133+ SK-N-BE we have (2) silenced cells, and en- observeddogenous that Elk-1 NANOG expression and in SOX2 CD133+ but SK-N-BE not POU5F1 (2) cells, were and downregulated observed that NANOGsignificantlyand uponSOX2 silencingbut not POU5F1 (Figure 4C).were It downregulatedmust be noted, however, significantly that uponoverexpression silencing (Figureof Elk-1-VP164C). It inmust CD133 be noted,− cells however,did not yield that overexpressionupregulation ofof NANOG Elk-1-VP16, SOX2 in CD133or POU5F1− cells in didSK-N-BE not yield (2) cellsupregulation (data not ofshown).NANOG , SOX2 or POU5F1 in SK-N-BE (2) cells (data not shown).

Figure 4. qPCRqPCR expression expression profiles of stemness genes in CD133− vs.vs. CD133+ CD133+ SK-N-BE(2) SK-N-BE(2) cells cells an andd in in primary primary brain brain tumors. tumors. (A). StemnessStemness gene gene expression expression analysis analysis of SKNBE(2)of SKNBE(2) passage passage 0, passage 0, passage 1, and 1, passage and passage 2 cells (***2 cellsp < (*** 0.001, p two-way< 0.001, two-way ANOVA w/DunnettANOVA w/Dunnett multiple multiple comparison comparison test); (B). test); Stemness (B). Stemness gene expression gene expression analysis of anal SKNBE(2)ysis of SKNBE(2) CD133+ BTICs CD133+ vs. BTICs CD133 vs.− CD133spheroids− spheroids (unpaired (unpaired t-test, * p t-test,< 0.05, * ***p

J. Pers. Med. 2021, 11, 125 19 of 28

To investigate whether similar regulation could be observed in primary GBM, primary brain tumor samples from three different patients (patient no. 428, 458, 624) were analyzed for ELK-1 expression in CD133− vs. CD133+ cells. Although there was variability between samples, in all three GBMs, CD133+ cells expressed significantly more ELK-1 than CD133− cells (Figure4D). This was parallel to our analysis of GBM cell lines, where tumorspheres of A172, T98G, and U87 GBM cells expressed significantly more Elk-1 protein than the monolayer cultures did, whereas ELK-1 expression level did not alter significantly in SH-SY5Y tumorsphere vs. monolayer cultures (data not shown). Furthermore, when endogenous ELK-1 was silenced in the primary tumor culture of patient 624 (middle level of ELK-1 expression), CD133, NANOG, SOX2, and POU5F1 levels were all downregulated as compared to scramble RNA control (Figure4E).

3.4. Effect of Elk-1 Expression on Colony Formation of SK-N-BE (2) Cells on Soft Agar The ability of transformed cells to grow in anchorage-free conditions is one of the hallmarks of cancer formation, and soft agar colony assay is a commonly used tool to assay for this feature [30]. It was shown in endometrial tumors, for instance, that CD133+ cells exhibited higher colony formation than CD133− cells in soft agar assay [31]. We have, therefore, addressed whether the same scenario was true for CD133+/CD133− SK-N-BE (2) cells, and whether overexpression of Elk-1-VP16 or silencing of endogenous Elk-1 would affect the number of colonies. To that end, we have sorted SK-N-BE (2) cells into CD133+ BTICs and CD133− cells, and both CD133+ and CD133− spheroids were grown in IPM culture conditions for three days in limiting dilution assay (LDA), and the frequency of spheroid formation was found to be almost tenfold more in CD133+ BTIC cells, indicating that sorting of cells was successful. Next, the effects of Elk-1 overexpression or silencing were studied; to that end, we have transfected CD133− cells with Elk-1-VP16 expression vector, while CD133+ cells were transfected with siElk-1 silencing vector as described in Materials and Meth- ods, and colony formation frequencies were determined in soft agar assay. In untrans- fected SK-N-BE (2) cells, CD133− cells and unsorted cells showed a similar number of colonies (24 ± 11 vs. 25 ± 11, respectively), whereas CD133+ BTICs had almost 50% more colonies formed (33 ± 9 colonies). CD133− cells transfected with either pCDNA3- Elk-1-VP16 (37 ± 10 colonies) or pCMV6-Flag-Elk-1-VP16 (50 ± 15 colonies) showed higher colony number than their counterparts transfected with empty vectors, pCDNA3.1 (32 ± 3 colonies) or pCMV-Flag (24 ± 10 colonies). On the other hand, CD133+ cells where endogenous Elk-1 was silenced by RNAi exhibited a decreased colony number (18 ± 5) compared to scrambled RNA control (26 ± 9) (Supplement Table S4).

3.5. Regulation of NANOG, POU5F1, and SOX2 Promoters by Elk-1 To assess whether the regulation of these genes by Elk-1 was direct or indirect, the promoters for NANOG, POU5F1, and SOX2 were cloned to luciferase reporter vectors and tested for Elk-1 regulation in different cell lines. Initially, SK-N-BE (2) (Figure5A) and SH-SY5Y (Figure5B) neuroblastoma cells and U87-MG (Figure5C), A172 (Figure5D), and T98G (Figure5E) GBM cells were either transfected with constitutively active Elk-1-VP16 and/or dominant-negative Elk-1-EN fusion protein expression vectors for overexpression (i), or with siElk-1 or scrRNA vectors for silencing (ii) experiments to study the regulation of SOX2 promoter by Elk-1 protein (Figure5). Although there appear to be cell type-specific variations, SOX2 promoter appeared to be upregulated upon Elk-1-VP16 overexpression in all cell types (Figure5Ai– Ei), whereas only SH-SY5Y and U87 cells exhibited downregulation of SOX2-dependent luciferase activity upon the silencing of endogenous Elk-1, indicating that other proteins are involved in the regulation of this promoter (Figure5Bii,Cii). J. Pers. Med. 2021, 11, 125 20 of 27 J. Pers. Med. 2021, 11, 125 20 of 28

Figure 5. SOX2 promoter activity analysis with respect to (i) Elk-1 variants over-expression and (ii) endogenous Elk-1 Figuresilencing 5. SOX2 in (A promoter) SK-NBE activity (2), (B) SH-SY5Y, analysis (withC) U87-MG, respect (toD )( A172,i) Elk-1 and variants (E) T98G over-expression cell lines. Luminometric and (ii) endogenous measurements Elk-1 silencingwere normalized in (A) SK-NBE to Renilla (2),-luc (B) activity. SH-SY5Y, ANOVA, (C) U87-MG, Tukey’s multiple(D) A172, comparative and (E) T98G tests, cell ** plines.< 0.01, Luminometric *** p < 0.001, **** measurementsp < 0.0001 werefor normalized Ai, Ci; unpaired to Renilla two-tailed-luc activity. t-test, * ANOVA,p < 0.5, ** pTukey’s< 0.01, ****multiplep < 0.0001 comparative was done tests, for Bi, ** Bii,p < 0.01, Cii, Di, *** Ei.p < 0.001, **** p < 0.0001 for Ai, Ci; unpaired two-tailed t-test, * p < 0.5, ** p < 0.01, **** p < 0.0001 was done for Bi, Bii, Cii, Di, Ei.

We next studied NANOG promoter; while SOX2 promoter was found to have 1 con- sensus Elk-1 binding motif with dissimilarity score (DS) of less than 1%, and 5 ets motifs

J. Pers. Med. 2021, 11, 125 21 of 27 J. Pers. Med. 2021, 11, 125 21 of 28

with DS 1–10%,We next NANOG studied NANOG promoterpromoter; was found while SOX2to containpromoter three was consensus found to have ets 1motifs (Fig- consensus Elk-1 binding motif with dissimilarity score (DS) of less than 1%, and 5 ets ure 6A),motifs two with of which DS 1–10%, hadNANOG DS of 1–10%promoter (T wasable found 2). We to contain have threeconstructed consensus a ets wildtype motifs NANOG promoter(Figure reporter6A), two vector of which (NANOG had DS of-Luc), 1–10% and (Table one2). Wewhere have the constructed higher asimilarity wildtype consensus ets motifNANOG (ets1)promoter was deleted reporter (NANOG vector (NANOG∆-Luc),-Luc), and and studied one where the the regu higherlation similarity of this promoter by Elk-1consensus in different ets motif cell (ets1) lines was (Figure deleted (6).NANOG ∆-Luc), and studied the regulation of this promoter by Elk-1 in different cell lines (Figure6).

Figure 6. Regulation of NANOG promoter by Elk-1. (A) Schematic diagram of predicted ets motifs ets1-3 on NANOG Figure 6. Regulationpromoter. Ets1of NANOGwas predicted promoter to be a strongerby Elk-1. binding (A) motifSchematic for Elk-1 diagram and was deletedof predicted to generate etsNANOG motifs∆ ets1-3-Luc reporter on NANOG pro- moter. Ets1 wasplasmid. predicted (B) Luciferase to be assaya stronger for (i) wildtype bindingNANOG motif-Luc for andElk-1 (ii )andNANOG was∆ -Lucdeleted reporters to generate in SK-N-BE NANOG (2) cells after∆-Luc reporter plasmid. (B) transfectionLuciferase of assay expression for ( plasmidsi) wildtype with Elk1-VP16,NANOG-Luc Elk1-EN, and or (ii empty) NANOG controlΔ plasmid-Luc reporters pCDNA3.1 in ( leftSK-N-BE graphs) or(2) cells after transfection ofco-transfection expression of plasmids silencing plasmids with Elk1-VP16, for scrRNA controlElk1-EN, or siElk-1 or empty (right control graphs). Luminometricplasmid pCDNA3.1 measurements (left were graphs) or co- transfection ofnormalized silencing to Renilla-lucplasmids activity.for scrRNA ANOVA, control Tukey’s or multiple siElk-1 comparative (right graphs tests, ***). Luminometricp < 0.001 for (i) and measurements (ii) left graphs; were nor- malized to Renilla-lucunpaired two-tailed activity. t-test, ANOVA, ** p < 0.01 wasTukey’s done formultiple (i) and (ii )comparative right graphs. (C tests,) Luciferase *** p assay < 0.001 for (i )for wildtype (i) andNANOG (ii) left-Luc graphs; un- paired two-tailedand (ii t-test,) NANOG ** ∆p-Luc < 0.01 reporters was done in SH-SY5Y for (i) cells and after (ii) transfectionright graphs. of expression (C) Luciferase plasmids assay with Elk1-VP16, for (i) wildtype Elk1-EN orNANOG-Luc empty control plasmid pCDNA3.1 (left graphs) or co-transfection of silencing plasmids for scrRNA control or siElk-1 and (ii) NANOGΔ-Luc reporters in SH-SY5Y cells after transfection of expression plasmids with Elk1-VP16, Elk1-EN or (right graphs). Luminometric measurements were normalized to Renilla-luc activity. ANOVA, Tukey’s multiple comparative empty controltests, plasmid *** p < 0.001 pCDNA3.1 for (i) and (iileft) left graphs graphs;) unpaired or co-transfection two-tailed t-test; of *silencingp < 0.5, *** pplasmids< 0.001 for (fori) and scrRNA (ii) right control graphs. or siElk-1 (right graphs(D).) Luminometric Luciferase assay formeasurements (i) wildtype NANOG were-Luc normalized and (ii) NANOG to Renilla∆-Luc-luc reporters activity. in U87-MG ANOVA, cells Tukey’s after transfection multiple compar- ative tests, ***of p expression < 0.001 plasmidsfor (i) and with ( Elk1-VP16,ii) left graphs; Elk1-EN, unpaired or empty two-tailed control plasmid t-test; pCDNA3.1 * p < 0.5, (left *** graphs p <) or0.001 co-transfection for (i) and (ii) right graphs. (D) Luciferaseof silencing plasmidsassay for for (i) scrRNA wildtype control NANOG or siElk-1-Luc (right and graphs (ii) NANOG). LuminometricΔ-Luc measurementsreporters in U87-MG were normalized cells after to transfec- tion of expressionRenilla -lucplasmids activity. with ANOVA, Elk1-VP16, Tukey’s multiple Elk1-EN, comparative or empty tests, control ** p < 0.01, plasmid *** p < 0.001 pCDNA3.1 for (i) and ((iileft) left graphs graphs; unpaired) or co-transfection of silencing plasmidst-test; ** p < for 0.01 scrRNA for (i) and control (ii) right graphs.or siElk-1 (right graphs). Luminometric measurements were normalized to Re- nilla-luc activity. ANOVA, Tukey’s multiple comparative tests, ** p < 0.01, *** p < 0.001 for (i) and (ii) left graphs; unpaired t-test; ** p < 0.01 for (i) and (ii) right graphs.

Elk-1-VP16 overexpression in SK-N-BE (2) cells resulted in upregulation from wildtype NANOG promoter, but the upregulation was slightly less in NANOG∆-Luc re- porter; Elk-1-EN repressed both promoter activities to control levels (Figure 6Bi vs. Figure 6Bii). Parallel to this, when Elk-1 was silenced using siElk-1, NANOG-Luc reporter activity was decreased (Figure 6Bi), whereas there was no significant change in NANOG∆-Luc activity in SK-N-BE (2) cells (Figure 6Bii). On the other hand, there was no significant difference between NANOG vs. NANOG∆ promoter activity by Elk-1-VP16 overexpres- sion in SH-SY5Y or U87 cells, while Elk-1-EN repressed both wildtype and mutant pro- moter activities (Figure 6C,D). There was a slight albeit significant increase in NANOG promoter activity in siElk-1 SH-SY5Y cells, whereas NANOG∆ promoter activity was de- creased upon siElk-1 silencing (Figure 6Ci vs. Figure 6Cii); in U87 silencing, endogenous

J. Pers. Med. 2021, 11, 125 22 of 28

Elk-1-VP16 overexpression in SK-N-BE (2) cells resulted in upregulation from wildtype NANOG promoter, but the upregulation was slightly less in NANOG∆-Luc reporter; Elk- 1-EN repressed both promoter activities to control levels (Figure6Bi vs. Figure6Bii). Parallel to this, when Elk-1 was silenced using siElk-1, NANOG-Luc reporter activity was decreased (Figure6Bi), whereas there was no significant change in NANOG ∆-Luc activity in SK-N-BE (2) cells (Figure6Bii). On the other hand, there was no significant difference between NANOG vs. NANOG∆ promoter activity by Elk-1-VP16 overexpression in SH-SY5Y or U87 cells, while Elk-1-EN repressed both wildtype and mutant promoter activities (Figure6 C,D). There was a slight albeit significant increase in NANOG promoter activity in siElk-1 SH-SY5Y cells, whereas NANOG∆ promoter activity was decreased upon siElk-1 silencing (Figure6Ci vs. Figure6Cii); in U87 silencing, endogenous Elk-1 did not significantly alter wildtype NANOG-Luc activity but resulted in a decrease in NANOG∆- Luc (Figure6Di vs. Figure6Dii). There was no significant change in either Elk-1-VP16 overexpression or siElk-1 silencing in A172 and T98G cells (data not shown). In POU5F1 promoter, of the four predicted ets motifs, three of them were predicted to have DS score of 1-10% DS (Table2; Figure7A). Wildtype POU5F1 promoter was cloned, and deletion constructs for these motifs (ets1-ets3) were generated for luciferase reporter assays as described in Materials and Methods. When wildtype POU5F1 promoter activity was compared to deletion constructs in SK-N-BE (2) cells transfected with Elk-1- VP16 expression plasmid, POU5F1∆ets2-Luc deletion construct exhibited less upregulation (around 2.4 units) than wildtype, POU5F1∆ets1, and POU5F1∆ets3 promoters (around 3 units), while Elk-1-EN overexpression resulted in similar level of activation to control in all cases (Figure7B). On the other hand, siElk-1 silencing did not result in a significant change in wildtype POU5F1 promoter activity or the POU5F1∆ets3 deletion mutant, whereas it resulted in a decrease in luciferase activity in both POU5F1∆ets1 and POU5F1∆ets2 constructs in SK-N-BE (2) cells (Figure7B). Elk-1-VP16 overexpression upregulated wildtype POU5F1-Luc reporter activity, while Elk-1-EN repressed it in SH-SY5Y cells; there was no significant change in this profile in either of the three ets deletion constructs, indicating the regulation might be through a different motif or could be indirect (Figure7C). Interestingly, siElk-1 silencing upreg- ulated wildtype POU5F1-Luc and POU5F1∆ets1-Luc reporter activity, while decreasing POU5F1∆ets2-Luc and POU5F1∆ets3-Luc reporter activity (Figure7C). In U87-MG GBM cells, however, wildtype POU5F1-Luc and POU5F1∆ets2-Luc reporters were upregulated to similar levels in Elk-1-VP16 overexpression (1.2 units in Figure7Di and 1.5 units in Figure7Diii), while POU5F1∆ets1-Luc was upregulated more (2.4 units, Figure7Dii), and upregulation was significantly less in POU5F1∆ets3-Luc reporter (Figure7Div). Elk-1-EN overexpression did not significantly alter promoter activity (Figure7D), and while siElk-1 silencing did not cause any change in wildtype promoter, it resulted in a downregulation in all deletion constructs to a different extent (Figure7D). Wildtype POU5F1-Luc promoter was upregulated by Elk-1-VP16 overexpression in both A172 and T98G cells, although siElk-1 silencing did not significantly change with respect to scrambled RNA control. J. Pers. Med. 2021, 11, 125 22 of 27

Elk-1 did not significantly alter wildtype NANOG-Luc activity but resulted in a decrease in NANOG∆-Luc (Figure 6Di vs. Figure 6Dii). There was no significant change in either Elk-1-VP16 overexpression or siElk-1 silencing in A172 and T98G cells (data not shown). In POU5F1 promoter, of the four predicted ets motifs, three of them were predicted to have DS score of 1-10% DS (Table 2; Figure 7A). Wildtype POU5F1 promoter was cloned, and deletion constructs for these motifs (ets1-ets3) were generated for luciferase reporter assays as described in Materials and Methods. When wildtype POU5F1 promoter activity was compared to deletion constructs in SK-N-BE (2) cells transfected with Elk-1- VP16 expression plasmid, POU5F1∆ets2-Luc deletion construct exhibited less upregula- tion (around 2.4 units) than wildtype, POU5F1∆ets1, and POU5F1∆ets3 promoters (around 3 units), while Elk-1-EN overexpression resulted in similar level of activation to control in all cases (Figure 7B). On the other hand, siElk-1 silencing did not result in a significant change in wildtype POU5F1 promoter activity or the POU5F1∆ets3 deletion J. Pers. Med. 2021, 11, 125 mutant, whereas it resulted in a decrease in luciferase activity in both POU5F1∆ets123 and of 28 POU5F1∆ets2 constructs in SK-N-BE (2) cells (Figure 7B).

FigureFigure 7. 7.Regulation Regulation of POU5F1 promoterpromoter by by Elk-1. Elk-1. (A ()A Schematic) Schematic diagram diagram of predicted of predicted ets motifsets motifs ets1-4ets1-4 on POU5F1on POU5F1 pro- promoter.moter. Motifs Motifs ets1-3 ets1-3 werewere predicted predicted to to be be stronger stronger binding binding motifs motifs for for Elk-1 Elk-1 and and were were individually individually deleted deleted to to generate generate POU5F1POU5F1∆∆eets1-ts1-Luc,Luc, POU5F1POU5F1∆∆eets2-ts2-Luc,Luc, andand POU5F1POU5F1∆ets3-Luc reporter plasmids. ( (BB) )Luciferase Luciferase assay assay for for ( (ii)) wildtype wildtype POU5F1-POU5F1-LucLuc and and itsits deletion mutant mutant reporters reporters (ii (ii) POU5F1) POU5F1∆e∆ts1-ets1-Luc,Luc, (iii ()iii POU5F1) POU5F1∆ets2-∆ets2-Luc,Luc, and and(iv) POU5F1 (iv) POU5F1∆ets3-∆Lucets3- Lucin inSK-N-BE SK-N-BE (2) (2)cells cells after after transfection transfection of ofexpression expression plasmids plasmids wi withth Elk1-VP16, Elk1-VP16, Elk1-EN, Elk1-EN, or or empty empty control control plasmid plasmid pCDNA3.1 (left graphs) or co-transfection of silencing plasmids for scrRNA control or siElk-1 (right graphs). Lumino- pCDNA3.1 (left graphs) or co-transfection of silencing plasmids for scrRNA control or siElk-1 (right graphs). Luminometric metric measurements were normalized to Renilla-luc activity. ANOVA, Tukey’s multiple comparative tests, ** p < 0.01, *** p p measurementsp < 0.001 for left were graphs normalized (i–iv); unpaired to Renilla-luc two-tailed activity. t-test, ANOVA, ** p < Tukey’s0.01 and multiple *** p < 0.001 comparative for right tests,graphs ** (i–

POU5F1∆ets3-Luc in U87-MG cells after transfection of expression plasmids with Elk1-VP16, Elk1-EN, or empty control plasmid pCDNA3.1 (left graphs) or co-transfection of silencing plasmids for scrRNA control or siElk-1 (right graphs). Luminometric measurements were normalized to Renilla-luc activity. ANOVA, Tukey’s multiple comparative tests, * p < 0.5, ** p < 0.01, *** p < 0.001 for left graphs (i–iv); unpaired two-tailed t-test, ** p < 0.01, *** p < 0.001 for right graphs (i–iv).

3.6. Binding of Elk-1 to Predicted ets Motifs on SOX2, NANOG, and POU5F1 Promoters Elk-1-VP16 overexpression was found to upregulate expression of SOX2, NANOG, and POU5F1 expression in qPCR analysis, and wildtype promoter luciferase reporters were found to be upregulated by Elk-1-VP16 in a cell context-dependent manner, yet deletion of predicted ets motifs did not significantly change reporter activities, indicating that either there are other ets motifs in distal promoters that are not cloned in this study, or that the regulation is not through direct Elk-1 binding to these predicted ets motifs. To address this second point, we have carried out chromatin immunoprecipitation (ChIP) experiments in SK-N-BE (2) neuroblastoma and T98G GBM cell lines (Figure8). J. Pers. Med. 2021, 11, 125 24 of 28

The cells were transfected with pCMV-Flag-Elk-1 (empty pCMV was used as control), and immunoprecipitation was carried out using Flag agarose beads (Flag IP); IgG beads were used as control (IgG IP). Elk-1 binding motifs on SRF and MCL-1 promoters were used as a positive control for Elk-1 binding. All three of the predicted ets motifs on the NANOG promoter exhibited Elk-1 binding in SK-N-BE (2) cells (Figure8A) but not on T98G cells (Figure8B). Similarly, all four predicted ets motifs on POU5F1 promoter showed Elk-1 binding, albeit to different extents, in SK-N-BE (2) cells (Figure8A) but not on T98G cells (Figure8B). Likewise, all five predicted ets motifs showed Elk-1 binding in SK-N-BE (2) cells (Figure8A), whereas only the ets3 motif showed significant binding to Elk-1 in T98G (Figure8B). This indicates that, while Elk-1 is capable of binding to these predicted motifs, this binding is affected by J. Pers. Med. 2021, 11, 125cell-dependent circumstances, which may be a transcriptional partner or posttranslational24 of 27

modification status of the Elk-1 protein in that particular cell type.

Figure 8. Chromatin immunoprecipitation assay for the identification of Elk-1 binding sites on the Figure 8. Chromatintarget immunoprecipitation gene promoters inassay pCMV-transfected for the identification (pCMV) of Elk-1 binding vs. Elk-1 sites over-expressing on the target gene cells promoters (Elk-1) in (A). in pCMV-transfected (pCMV) vs. Elk-1 over-expressing cells (Elk-1) in (A). SK-N-BE (2) cells and (B). T98G cells. Lysates were immunoprecipitatedSK-N-BE with (2) cellseither and Flag (antibodyB). T98G (Flag cells. IP) Lysates for exogenous were Elk-1 immunoprecipitated or IgG (IgG IP) as control. with eitherqPCR results Flag antibody were analyzed as(Flag explained IP) for in exogenous Materials and Elk-1 Methods or IgG and (IgG reported IP) as as control. average qPCRfold change. results All were predicted analyzed ets motif as explained se- in quences on NANOGMaterials, SOX2, and POU5F1 Methods promoters and reported were screened as average for Elk-1 fold binding; change. SRF All and predicted MCL1 promoterets motif sequences sequences on were used as positive control. NANOG, SOX2, and POU5F1 promoters were screened for Elk-1 binding; SRF and MCL1 promoter sequences4. were Discussion used as positive control. ETS transcription factors are involved in a number of biological processes in different tissues, and it was shown that in embryonic development, expression of several ETS pro- teins including Elf3 and SpiC increased after fertilization until the blastocyst stage, and silencing of ETS expression affected Oct3/4 gene expression [32]. It was shown in human pluripotent stem cells (hPSCs) with different X inactivation states (Xa, active, Xi, inactive) that Elk-1 overexpression mimicked XaXa in terms of decreased pluripotency, the differences being diminished in low oxygen [33].

J. Pers. Med. 2021, 11, 125 25 of 28

4. Discussion ETS transcription factors are involved in a number of biological processes in different tissues, and it was shown that in embryonic development, expression of several ETS proteins including Elf3 and SpiC increased after fertilization until the blastocyst stage, and silencing of ETS expression affected Oct3/4 gene expression [32]. It was shown in human pluripotent stem cells (hPSCs) with different inactivation states (Xa, active, Xi, inactive) that Elk-1 overexpression mimicked XaXa in terms of decreased pluripotency, the differences being diminished in low oxygen [33]. One study has shown Elk-1 to be essential for human embryonic stem cells, and that it co-occupies promoters of genes in cell proliferation pathways with ERK2, and in the absence of ERK2, the promoters were repressed by Polycomb proteins [29]. In fact, Elk-1 was further found to be upregulated, while Nanog, Oct4, and Sox2 were found to be repressed during mesoderm differentiation of hESCs, and it was shown to bind to and activate promoters such as EGR-1 while repressing a subset of promoters such as FOSL1 [16]. Intriguingly, mice deficient for Elk-1 were viable albeit with mild neuronal impairment, indicating other Ets proteins may act redundantly and compensate for its embryonic functions [34]. During neuronal differentiation of mES cells, Sox2 chromatin interaction profiles were altered, and promoters of neuronal differentially expressed gene clusters were enriched in Elk-1, among other transcription factors [35]. Similarly, during reprogramming of fibroblasts into neural stem cells (NSCs) using pharmacological molecules, Elk-1 was found to be one of the transcription factors to regulate reprogramming, particularly through binding Sox2 promoter [36]. Another Ets protein, Pea3/ETV4, was shown to regulate Nanog and Oct4 expression in pluripotent NCCIT embryonic carcinoma cells [37,38]. Interestingly, members of the Pea3 subfamily of ETS proteins, ETV4 and ETV5, were found to be expressed in undif- ferentiated ES cells, and suppression of Oct3/4 was found to result in downregulation of their expression, and ETV4 and ETV5 were found to be important for proliferation of undifferentiated ES cells through regulation of stem cell-related genes such as Tcf15, Gbx2, and Zic3 [39]. A transcriptional partner of Elk-1, namely serum response factor (SRF), was shown to repress the reprogramming induced by ERK pathway inhibition, and to negatively regulate pluripotency [40], which may be independent of Elk-1 interaction. CD133 is a cell surface protein that has been used alone [12] or in combination with CD15 [41] to isolate and culture brain-tumor-initiating cells from a variety of tumors. ERK/MAPK pathway was shown to be required for CD133 expression [42], and HIF-1α was shown to bind to the CD133 promoter through Elk-1 [15], which is supported in our study by overexpression of Elk-1 in CD133+ BTIC subpopulation. In a genome-wide study in the human embryonic stem cell (hESC) population, ELK1 was found to be essential for hESCs, and some of the promoters bound by ELK1 were determined to be important in the maintenance of embryonic identity, spinal cord development, and neuron fate development [29]. Furthermore, induced neural stem cells were found to contain relatively high levels of phosphorylated Elk-1, along with Gli2, and both were shown to bind to Sox2 promoter upon neural reprogramming [36], and distinct GABPA/Elk-1 motifs were found in Sox2 promoter, identified as a neuronal cluster gene involved in differentiation of embryonic stem cells to neuronal precur- sors [35]. It is intriguing whether tumorigenesis reactivates this mechanism in a cell context-dependent manner.

5. Conclusions We propose that not only does ELK1 present a novel target for tumor therapy directed at eliminating BTIC population, but also can be used as a molecular diagnostic molecule to identify potential for tumor recurrence. It should be noted, however, that posttranslational modifications such as phosphorylation and SUMOylation regulate ELK1 protein, which can differ among gliomas and must be studied in more detail. J. Pers. Med. 2021, 11, 125 26 of 28

Supplementary Materials: The following are available online at https://www.mdpi.com/xxx/s1, supppl1: Biological processes among enriched pathways that were upregulated in Elk-1-VP16- transfected SH-SY5Y cells, Table S2: Biological processes among enriched pathways that were downregulated in Elk-1-VP16-transfected SH-SY5Y cells, Table S3: Transcription factor binding site analysis for promoters of genes identified in microarray analysis, Table S4: Soft agar assay colony formation assay. Author Contributions: Conceptualization, I.A.K.; data curation, U.D.; formal analysis, M.S.S., C.V., B.K., U.D., S.M., G.G., and I.A.K.; funding acquisition, I.A.K.; investigation, M.S.S., C.V., B.K., U.D., and S.S.; methodology, M.S.S., B.K., S.M., S.S., G.G., and K.Y.A.; project administration, B.Y. and I.A.K.; resources, S.S. and I.A.K.; software, G.G.; supervision, K.Y.A., B.Y., and I.A.K.; writing—original draft, M.S.S., C.V., B.K., and S.S.; writing—review and editing, M.S.S., C.V., S.S., K.Y.A., and B.Y. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by TUBITAK project no 211T167 as part of the COST TD901 Action, and TUBITAK project no 115Z804 as part of COST TN1301 Action. Institutional Review Board Statement: Human GBM samples were obtained from consenting patients, as approved by the Hamilton Health Sciences/McMaster Health Sciences Research Ethics Board (REB# 07-366; approved 2007, last updated September 2020). Informed Consent Statement: Not applicable. Data Availability Statement: Transcriptomics raw data have been uploaded to and available at EBI ArrayExpress, with accession number of E-MTAB-9938. The rest of the data are available upon request. Acknowledgments: We would like to thank members of the Kurnaz and Singh laboratories for helpful discussions and Eray Sahin for initial data analysis of the Elk-1-VP16 microarray with demo IPA software. IAK is a GYA member and GG is a YOK 100/2000 student. Conflicts of Interest: The authors declare no conflict of interest.

References 1. Davis, S.P.; Vanhoutte, C.; Pages, J. Caboche and S. Laroche. The MAPK/ERK cascade targets both Elk-1 and cAMP response element-binding protein to control long-term potentiation-dependent gene expression in the dentate gyrus in vivo. J. Neurosci. 2000, 20, 4563–4572. [CrossRef] 2. Demir, O.; Ari, O.; Kurnaz, I.A. Elk-1 interacts with dynein upon serum stimulation but independent of serine 383 phosphorylation. Cell. Mol. Neurobiol. 2012, 32, 185–189. [CrossRef] 3. Demir, O.; Korulu, S.; Yildiz, A.; Karabay, A.; Kurnaz, I.A. Elk-1 interacts with neuronal microtubules and relocalizes to the nucleus upon phosphorylation. Mol. Cell. Neurosci. 2009, 40, 111–119. [CrossRef] 4. Kaminska, B.; Kaczmarek, L.; Zangenehpour, S.; Chaudhuri, A. Rapid phosphorylation of Elk-1 transcription factor and activation of MAP kinase signal transduction pathways in response to visual stimulation. Mol. Cell. Neurosci. 1999, 13, 405–414. [CrossRef] 5. Sananbenesi, F.; Fischer, A.; Schrick, C.; Spiess, J.; Radulovic, J. Phosphorylation of hippocampal Erk-1/2, Elk-1, and p90-Rsk-1 during contextual fear conditioning: Interactions between Erk-1/2 and Elk-1. Mol. Cell. Neurosci. 2002, 21, 463–476. [CrossRef] 6. Sgambato, V.; Vanhoutte, P.; Pages, C.; Rogard, M.; Hipskind, R.; Besson, M.J.; Caboche, J. In vivo expression and regulation of Elk-1, a target of the extracellular-regulated kinase signaling pathway, in the adult rat brain. J. Neurosci. 1998, 18, 214–226. [CrossRef] 7. Yang, S.H.; Shore, P.; Willingham, N.; Lakey, J.H.; Sharrocks, A.D. The mechanism of phosphorylation-inducible activation of the ETS-domain transcription factor Elk-1. EMBO J. 1999, 18, 5666–5674. [CrossRef][PubMed] 8. Yang, S.H.; Whitmarsh, A.J.; Davis, R.J.; Sharrocks, A.D. Differential targeting of MAP kinases to the ETS-domain transcription factor Elk-1. EMBO J. 1998, 17, 1740–1749. [CrossRef] 9. Sharrocks, A.D. The ETS-domain transcription factor family. Nat. Rev. Mol. Cell Biol. 2001, 2, 827–837. [CrossRef][PubMed] 10. Boros, J.; Donaldson, I.J.; O’donnell, A.; Odrowaz, Z.A.; Zeef, L.; Lupien, M.; Meyer, C.A.; Liu, X.S.; Brown, M.; Sharrocks, A.D. Elucidation of the ELK1 target gene network reveals a role in the coordinate regulation of core components of the gene regulation machinery. Genome Res. 2009, 19, 1963–1973. [CrossRef] 11. Demir, O.; Aysit, N.; Onder, Z.; Turkel, N.; Ozturk, G.; Sharrocks, A.D.; Kurnaz, I.A. ETS-domain transcription factor Elk-1 mediates neuronal survival: SMN as a potential target. Biochim. Biophys. Acta 2011, 1812, 652–662. [CrossRef] 12. Lenkiewicz, M.; Li, N.; Singh, S.K. Culture and isolation of brain tumor initiating cells. Curr. Protoc. Stem Cell Biol. 2009, 11, 3.3.1–3.3.10. [CrossRef] J. Pers. Med. 2021, 11, 125 27 of 28

13. Mahller, Y.Y.; Williams, J.P.; Baird, W.H.; Mitton, B.; Grossheim, J.; Saeki, Y.; Cancelas, J.A.; Ratner, N.; Cripe, T.P. Neuroblastoma cell lines contain pluripotent tumor initiating cells that are susceptible to a targeted oncolytic virus. PLoS ONE 2009, 4, e4235. [CrossRef] 14. Singh, S.K.; Hawkins, C.; Clarke, I.D.; Squire, J.A.; Bayani, J.; Hide, T.; Henkelman, R.M.; Cusimano, M.D.; Dirks, P.B. Identification of human brain tumour initiating cells. Nature 2004, 432, 396–401. [CrossRef] 15. Ohnishi, S.; Maehara, O.; Nakagawa, K.; Kameya, A.; Otaki, K.; Fujita, H.; Higashi, R.; Takagi, K.; Asaka, M.; Sakamoto, N.; et al. hypoxia-inducible factors activate CD133 promoter through ETS family transcription factors. PLoS ONE 2013, 8, e66255. [CrossRef] 16. Prise, I.; Sharrocks, A.D. ELK1 has a dual activating and repressive role in human embryonic stem cells. Wellcome Open Res. 2019, 4, 41. [CrossRef] 17. Vora, P.; Seyfrid, M.; Venugopal, C.; Qazi, M.A.; Salim, S.; Isserlin, R.; Subapanditha, M.; O’farrell, E.; Mahendram, S.; Singh, M.; et al. Bmi1 regulates human glioblastoma stem cells through activation of differential gene networks in CD133+ brain tumor initiating cells. J. Neurooncol. 2019, 143, 417–428. [CrossRef] 18. Venugopal, C.; Hallett, R.; Vora, P.; Manoranjan, B.; Mahendram, S.; Qazi, M.A.; Mcfarlane, N.; Subapanditha, M.; Nolte, S.M.; Singh, M.; et al. Pyrvinium Targets CD133 in Human Glioblastoma Brain Tumor-Initiating Cells. Clin. Cancer Res. 2015, 21, 5324–5337. [CrossRef] [PubMed] 19. Venugopal, C.; Mcfarlane, N.M.; Nolte, S.; Manoranjan, B.; Singh, S.K. Processing of primary brain tumor tissue for stem cell assays and flow sorting. J. Vis. Exp. 2012.[CrossRef] 20. Kandemir, B.; Dag, U.; Bakir Gungor, B.; Durasi, I.M.; Erdogan, B.; Sahin, E.; Sezerman, U.; Aksan Kurnaz, I. In silico analyses and global transcriptional profiling reveal novel putative targets for Pea3 transcription factor related to its function in neurons. PLoS ONE 2017, 12, e0170585. [CrossRef] 21. Bolstad, B.M.; Irizarry, R.A.; Astrand, M.; Speed, T.P. A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 2003, 19, 185–193. [CrossRef] 22. Kamburov, A.; Stelzl, U.; Lehrach, H.; Herwig, R. The ConsensusPathDB interaction database: 2013 update. Nucleic Acids Res. 2013, 41, D793–D800. [CrossRef] 23. Kanehisa, M.; Sato, Y.; Kawashima, M.; Furumichi, M.; Tanabe, M. KEGG as a reference resource for gene and protein annotation. Nucleic Acids Res. 2016, 44, D457–D462. [CrossRef] 24. Fabregat, A.; Sidiropoulos, K.; Garapati, P.; Gillespie, M.; Hausmann, K.; Haw, R.; Jassal, B.; Jupe, S.; Korninger, F.; Mckay, S.; et al. The Reactome pathway Knowledgebase. Nucleic Acids Res. 2016, 44, D481–D487. [CrossRef] 25. Nishimura, A. BioCarta. Biotech. Softw. Internet Rep. 2001, 2, 117–120. [CrossRef] 26. Ashburner, M.; Ball, C.A.; Blake, J.A.; Botstein, D.; Butler, H.; Cherry, J.M.; Davis, A.P.; Dolinski, K.; Dwight, S.S.; Eppig, J.T.; et al. Gene ontology: Tool for the unification of biology. The Gene Ontology Consortium. Nat. Genet. 2000, 25, 25–29. [CrossRef] 27. Fishilevich, S.; Zimmerman, S.; Kohn, A.; Iny Stein, T.; Olender, T.; Kolker, E.; Safran, M.; Lancet, D. Genic insights from integrated human proteomics in GeneCards. Database 2016, 2016, baw030. [CrossRef][PubMed] 28. Messeguer, X.; Escudero, R.; Farre, D.; Nunez, O.; Martinez, J.; Alba, M.M. PROMO: Detection of known transcription regulatory elements using species-tailored searches. Bioinformatics 2002, 18, 333–334. [CrossRef][PubMed] 29. Goke, J.; Chan, Y.S.; Yan, J.; Vingron, M.; Ng, H.H. Genome-wide kinase-chromatin interactions reveal the regulatory network of ERK signaling in human embryonic stem cells. Mol. Cell 2013, 50, 844–855. [CrossRef] 30. Borowicz, S.; Van Scoyk, M.; Avasarala, S.; Karuppusamy Rathinam, M.K.; Tauler, J.; Bikkavilli, R.K.; Winn, R.A. The soft agar colony formation assay. J. Vis. Exp. 2014, e51998. [CrossRef] 31. Ding, D.C.; Liu, H.W.; Chang, Y.H.; Chu, T.Y. Expression of CD133 in endometrial cancer cells and its implications. J. Cancer 2017, 8, 2142–2153. [CrossRef] 32. Kageyama, S.; Liu, H.; Nagata, M.; Aoki, F. The role of ETS transcription factors in transcription and development of mouse preimplantation embryos. Biochem. Biophys. Res. Commun. 2006, 344, 675–679. [CrossRef] 33. Bruck, T.; Yanuka, O.; Benvenisty, N. Human pluripotent stem cells with distinct X inactivation status show molecular and cellular differences controlled by the X-Linked ELK-1 gene. Cell Rep. 2013, 4, 262–270. [CrossRef] 34. Cesari, F.; Brecht, S.; Vintersten, K.; Vuong, L.G.; Hofmann, M.; Klingel, K.; Schnorr, J.J.; Arsenian, S.; Schild, H.; Herdegen, T.; et al. Mice deficient for the ets transcription factor elk-1 show normal immune responses and mildly impaired neuronal gene activation. Mol. Cell. Biol. 2004, 24, 294–305. [CrossRef] 35. Bunina, D.; Abazova, N.; Diaz, N.; Noh, K.M.; Krijgsveld, J.; Zaugg, J.B. Genomic Rewiring of SOX2 Chromatin Interaction Network during Differentiation of ESCs to Postmitotic Neurons. Cell Syst. 2020, 10, 480–494.e8. [CrossRef] 36. Zhang, L.; Gong, Y.; Zhao, X.; Zhou, H. Comparative study between hypoxia and hypoxia mimetic agents on osteogenesis of bone marrow mesenchymal stem cells in mouse. Zhongguo Xiu Fu Chong Jian Wai Ke Za Zhi 2016, 30, 903–908. 37. Park, S.W.; Do, H.J.; Choi, W.; Song, H.; Chung, H.J.; Kim, J.H. NANOG gene expression is regulated by the ETS transcription factor ETV4 in human embryonic carcinoma NCCIT cells. Biochem. Biophys. Res. Commun. 2017, 487, 532–538. [CrossRef] 38. Park, S.W.; Do, H.J.; Ha, W.T.; Han, M.H.; Park, K.H.; Song, H.; Kim, N.H.; Kim, J.H. Transcriptional activation of OCT4 by the ETS transcription factor PEA3 in NCCIT human embryonic carcinoma cells. FEBS Lett. 2014, 588, 3129–3136. [CrossRef] J. Pers. Med. 2021, 11, 125 28 of 28

39. Akagi, T.; Kuure, S.; Uranishi, K.; Koide, H.; Costantini, F.; Yokota, T. ETS-related transcription factors ETV4 and ETV5 are involved in proliferation and induction of differentiation-associated genes in embryonic stem(ES) cells. J. Biol. Chem. 2015, 290, 22460–22473. [CrossRef] 40. Huh, S.; Song, H.R.; Jeong, G.R.; Jang, H.; Seo, N.H.; Lee, J.H.; Yi, J.Y.; Lee, B.; Choi, H.W.; Do, J.T.; et al. Suppression of the ERK-SRF axis facilitates somatic cell reprogramming. Exp. Mol. Med. 2018, 50, e448. [CrossRef] 41. Vora, P.; Venugopal, C.; Mcfarlane, N.; Singh, S.K. Culture and Isolation of Brain Tumor Initiating Cells. Curr. Protoc. Stem Cell Biol. 2015, 34, 3.3.1–3.3.13. [CrossRef] 42. Tabu, K.; Kimura, T.; Sasai, K.; Wang, L.; Bizen, N.; Nishihara, H.; Taga, T.; Tanaka, S. Analysis of an alternative human CD133 promoter reveals the implication of Ras/ERK pathway in tumor stem-like hallmarks. Mol. Cancer 2010, 9, 39. [CrossRef]